We have looked at numerous examples of the time it took recently on other tickers to be processed. It should be tomorrow morning. And that is a very exciting thing.

SECURE ARTIFICIAL INTELLIGENCE ENABLED MEDICAL SENSOR PLATFORMS
A secure artificial intelligence (AI) enabled wearable medical sensor platform is used for adaptive operation according to features and techniques described herein. Operational parameters are modified based on data inputs thereto that provide feedback to the AI systems of the wearable sensor platform. The described technology can facilitate adaptive optimizations provided by AI machine learning algorithms in a manner that can beneficially assist in the monitoring and treatment of a patient. For example, the system described herein may be used for the continuous monitoring of the physiological parameters and health of a patient's vascular access point (for example, the fistula) and may provide, among other things, early warnings of possible infection at the vascular access location.


This application is a continuation application of and claims the benefit of priority to U.S. application Ser. No. 16/897,896, filed on Jun. 10, 2020, the contents of which are hereby incorporated by reference.

This disclosure generally relates to sensor platforms for monitoring the health of a user including secure artificial intelligence (AI) enabled wearable sensor platforms.

Patients with kidney failure or partial kidney failure typically undergo dialysis treatment using medical devices such as dialysis machines. The two principal dialysis methods are hemodialysis (HD) and peritoneal dialysis (PD). During hemodialysis, the patient's blood is passed through a dialyzer of a hemodialysis machine while also passing dialysate through the dialyzer. A semi-permeable membrane in the dialyzer separates the blood from the dialysate within the dialyzer and allows diffusion and osmosis exchanges to take place between the dialysate and the blood stream. Before the blood is returned to the body, air bubbles are removed from the blood to inhibit embolisms. The process of removing air is typically accomplished through use of a venous drip chamber, which is located downstream of the blood outlet of a dialyzer and upstream of the venous blood return of the patient.

During peritoneal dialysis, the patient's peritoneal cavity is periodically infused with dialysate, or dialysis solution. The membranous lining of the patient's peritoneum acts as a natural semi-permeable membrane that allows diffusion and osmosis exchanges to take place between the solution and the blood stream. Automated peritoneal dialysis machines, also called PD cyclers, are designed to control the entire peritoneal dialysis process so that it can be performed at home, usually overnight, without clinical staff in attendance. Both HD and PD machines may include displays with touch screens or other user interfaces that display information of a dialysis treatment and/or enable an operator or patient to interact with the machine.

When healthy, kidneys maintain the body's internal equilibrium of water and minerals (e.g., sodium, potassium, chloride, calcium, phosphorous, magnesium, and sulfate). In hemodialysis, blood is taken from a patient through an intake needle (or catheter) which draws blood from an artery located in a specific accepted access location (arm, thigh, subclavian, etc.). The drawn blood is pumped through extracorporeal tubing via a peristaltic pump, and then through a special filter called a dialyzer. The dialyzer is intended to remove unwanted toxins such as blood urea, nitrogen, potassium, and excess water from the blood. The dialyzed blood then flows out of the dialyzer via tubing and a needle (or catheter) back into the patient.

Vascular access is used for hemodialysis. These accesses are expected to sustain blood flow rates above 300 mL/min. One type of access is the arteriovenous fistula (AVF). AVF is considered to be the most effective of the vascular accesses for dialysis patients. One reason AVFs are effective is that they use native blood vessels, which, compared to synthetic grafts, are less likely to develop stenosis and fail. AVFs for hemodialysis typically include a surgically created connection between an artery and a vein in the arm. Artery walls are thicker than vein walls, and since the AVF connects the artery and the vein, the walls of the veins become thicker due to higher arterial pressure. The strengthened vein can then tolerate needles during hemodialysis.

Many different types of systems benefit from adaptive operation that is provided when used with artificial intelligence. In such systems, operational parameters are modified based on data inputs thereto that provide feedback to the artificial intelligence systems.

SUMMARY
In at least one aspect of the present disclosure a wearable sensor platform is provided. The wearable sensor platform includes: a plurality of sensors configured to capture raw sensor data; a transceiver; one or more storage devices storing computer-executable instructions; and one or more computer processors configured to execute the computer-executable instructions to cause the one or more computer processors to perform operations. The operations include: adjusting, using a machine learning model, one or more algorithms for processing the captured raw sensor data based on user data of a user of the wearable sensor platform; receiving the captured raw sensor data from the one or more sensors; determining, using the adjusted one or more algorithms and based on the captured raw sensor data, differential readings between pairs of sensors of the plurality of sensors; determining, using the adjusted one or more algorithms, whether the differential readings exceed a threshold; and generating an alert in response to determining that the differential readings exceed the threshold.

The operations can further include: receiving constraint data indicating one or more constraints corresponding to the wearable sensor platform; and applying the constraint data to the machine learning model during the adjusting. The operations can further include receiving, from an external source, one or more encryption keys; encrypting, use the one or more encryption keys, the captured raw sensor data; and transmitting the captured raw sensor data to the external source using the transceiver. The external source can include one or more processors of a dialysis system. The one or more encryption keys can include one or more multi-mode encoded keys. The threshold can be based on baseline reference data of the user learned by the machine learning model. The operations can further include: determining, using the adjusted one or more algorithms and based on the captured sensor data, whether one or more physiological measurements of the user deviate from a nominal value by a second threshold; and determining, in response to determining that one or more physiological measurements of the user deviated from the nominal value by the second threshold, trend information for the one or more physiological measurements, the trend information indicating a change in the one or more physiological measurements over time.

These and other aspects, features, and implementations can be expressed as methods, apparatus, systems, components, program products, methods of doing business, means or steps for performing a function, and in other ways, and will become apparent from the following descriptions, including the claims.

Implementations can include one or more of the following advantages. Continuous monitoring of the physiological parameters and health of a patient can be accomplished to provide, among other things, early warnings of possible problems. The described technology can facilitate adaptive optimizations provided by machine learning algorithms in a manner that can assist in providing a physiological profile tuned to a user's natural physical state while a sensor system retains validated baseline parameters and a baseline threshold derived from, for example, national standards. The described technology further provides for enhanced security related to the capture and use of a patient's data.

For example, the system described herein may be used for the continuous monitoring of the physiological parameters and health of a patient's vascular access point (for example, the fistula) and may provide, among other things, early warnings of possible infection at the vascular access location. Critical physical (corporal) characteristics near a patient's access point can be continuously monitored. The described technology can be used to increase cost saving in maintaining a vascular access point of the user. Further, the described technology can be used to increase the blood transfer efficiency through the vascular access point, and increase the lifespan and reduction in maintenance cost related to the vascular access point.

These and other aspects, features, and implementations will become apparent from the following descriptions, including the claims.

BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 shows a perspective view of a hemodialysis system, according to one or more implementations of the present disclosure.

FIG. 2 shows a perspective view of a peritoneal dialysis system, according to one or more implementations of the present disclosure.

FIG. 3A shows a block diagram of a secure AI enabled wearable sensor platform, according to one or more implementations of the present disclosure.

FIG. 3B shows a block diagram of an environment for monitoring a vascular access, according to one or more implementations of the present disclosure.

FIG. 4 shows a flowchart of a method for monitoring the health of a patient, according to one or more implementations of the present disclosure.

FIG. 5 shows a secure AI enabled wearable sensor platform having temperature sensors for differential temperature measurements, according to one or more implementations of the present disclosure.

FIG. 6 shows a secure AI enabled wearable sensor platform having temperature sensors for gradient temperature measurements, according to one or more implementations of the present disclosure.

FIG. 7 shows a secure AI enabled wearable sensor platform having pressure sensors, according to one or more implementations of the present disclosure.

FIG. 8 shows a secure AI enabled wearable sensor platform having spectral sensors, according to one or more implementations of the present disclosure.

FIG. 9 shows a block diagram of an environment including a factory module, a supervisor module, and a plurality of operations modules, according to one or more implementations of the present disclosure.

FIG. 10 shows a flowchart of a method performed in connection with transferring calibration and configuration data, baseline reference data, and security keys from a factory module to a supervisor module, according to one or more implementations of the present disclosure.

FIG. 11A shows a flowchart of a method performed by an AI-machine learning system in conjunction with a security enforcer, according to one or more implementations of the present disclosure.

FIG. 11B shows a flowchart of a method performed by a security enforcer in conjunction with an AI-machine learning system, according to one or more implementations of the present disclosure.

FIG. 12 shows a block diagram of an environment including an AI-machine learning system, a security engine, a dynamic operational data set, and a baseline reference data set, according to one or more implementations of the present disclosure.

FIG. 13 is a block diagram of an example computer system used to provide computational functionalities associated with described algorithms, methods, functions, processes, flows, and procedures described in the present disclosure.

FIG. 14 is a schematic diagram illustrating information flow and interaction between different components and modules of a wearable sensor platform.

FIG. 15 is a schematic diagram showing an example of a connected medical information distribution system, for example, a connected health (CH) system, that may that may be used in connection with network aspects of the wearable sensor platform described herein.

DETAILED DESCRIPTION
A secure AI enabled wearable medical sensor platform may be used for adaptive operation according to features and techniques described herein. Operational parameters are modified based on data inputs thereto that provide feedback to the AI systems of the wearable sensor platform. The described technology can facilitate adaptive optimizations provided by AI machine learning algorithms in a manner that can beneficially assist in the monitoring and treatment of a patient. For example, the system described herein may be used for the continuous monitoring of the physiological parameters and health of a patient's vascular access point (for example, the fistula) and may provide, among other things, early warnings of possible infection at the vascular access location.

Although AVFs are effective accesses for patients undergoing hemodialysis, there are constraints that lead AVFs to fail. AVFs can become infected because they are exposed sites that facilitate removing blood from patients and reintroducing blood to the patients. Patients undergo hemodialysis about three times a week so there are multiple opportunities during the week for the AVF to become infected. Furthermore, AVFs can fail due to stenosis, improper maintenance during maturation period, hematoma, and a patient's overall lack of awareness. Typically, current AVF monitoring technologies are not wearable and do not perform continuous monitoring functions. Also, current AVF monitoring technologies can be invasive and are usually performed once every four weeks (and/or while a patient is receiving dialysis treatment). In these technologies, blood is recirculated, and an ultrasound or ultrasonic sensor is used to monitor the blood recirculation to determine whether the AVF is performing well. These technologies may also need blood chemistry lab results to measure various chemical and biological parameters.

Furthermore, the effectiveness of dialysis can depend on the patient's chemical, biological, and physical characteristics. A relatively healthy kidney may include natural biological sensory that can facilitate the maintenance of a patient's corporeal wellbeing, such as blood pressure control and production of hemoglobin. However, in a patient with chronic kidney disease (CKD), many of these natural biological sensors can become ineffective, and the corresponding natural corporeal sensory feedback control mechanisms can become less effective. For example, the lack of effective natural sensors embedded in the kidney can introduce unpredictable and/or unmanageable changes in the state of natural health of the patient (e.g., deficiency in blood pressure control, deficiency in the rate of hemoglobin production, and so forth). Furthermore, in systems that implement AI-machine learning, operational parameters can be modified based on sensor data inputs that provide feedback to the machine learning systems. For example, a speech recognition system may adapt and improve based on data input that indicates successes and failures. The adaptive optimizations may be provided by artificial intelligence processing that revises operation states of the speech recognition system. However, the adaptive optimizations may cause the systems to operate outside of design parameters that impact safety considerations. Also, raw patient data may include sensitive information, and the collection and distribution of such data may raise privacy concerns.

Implementations of the present disclosure can provide means to alleviate some of the aforementioned disadvantages. Some implementations of the present disclosure provide a secure AI enabled wearable sensor platform designed to utilize different sensors to detect onset of infection growth, clots from stenosis, and changes in the local area occurring due to improper maintenance. Some implementations of the present disclosure provide subjective assessments for whether an AVF is fine or whether an AVF is at risk. Some implementations of the present disclosure alert a clinic or an emergency contact to intervene when the AVF is at risk. As such, the intervention can come at a timely fashion to prevent further damage to the AVF. The patient's AVF can, therefore, be monitored more frequently, even continuously, as opposed to every four weeks in conventional methods.

Implementations of the present disclosure can provide systems and devices (for example, a secure AI enabled wearable sensor platform) that use machine learning to continuously track and monitor one or more sensors distributed near a patient's fistula. In some implementations, the described systems and devices use an AI assisted feedback control system (sometimes referred to as an analog front end in this specification) to optimize sensor response characteristics for a specific target objective. In some implementations, the described systems and devices use an AI assisted sensor sampling rate to optimize power usage for specific operational goals, track and monitor sensor performance, and track and monitor calibration data. In some implementations, the described systems and methods use AI assisted signal processing to provide suggestions for a medical professional (for example, a physician overseeing dialysis treatment) that may be used by the medical professional to optimize treatment prescription data. In some implementations, the described systems and devices can perform AI assisted system root-cause analysis and/or determine sensor degradation status, and can include supervisory features to limit AI control features to be within original factory validated/acceptance states.

The types of sensors used can be mapped to the natural biological sensors that may have become degraded or ineffective due to physiological/medical abnormalities. For example, the one or more sensors can include, among others, one or more of the following: a temperature sensor, pulse oximetry (SpO2) sensor, blood pressure sensor, blood flow rate sensor, phosphorous sensors, sodium sensors, and potassium sensors. Implementations of the described systems and devices can periodically determine the health of the access point (fistula) by determining one or more health parameters associated with the fistula, such as optimal transfer rate, likelihood of infection, and so forth. Implementations of the described systems and devices can determine one or more health parameters associated with a natural balanced corporeal operational state of the fistula that may have a vital impact on dialysis transfer efficiency (for example, natural body temperature, blood pressure, SpO2, erythropoietin level, blood flow rate, and so forth). Implementations of the present disclosure can determine the corporal operational state while the patient is undergoing dialysis, or when the patient is not undergoing dialysis.

Implementations of the present disclosure can use secure hash algorithms (SHAs), such as SHA-256 algorithms, to provide secure key based modification (and/or ability to enable/disable) of operational/functional features of AI supervisory functions, calibration algorithms, data/signal processing algorithms, and boundary conditions for the machine learning system to maintain deterministic operational stability of the entire platform. SHA-256 based multimode encoded key algorithms can be used to authenticate each device to a unique patient, patient's treatment data with respect to each dialysis treatment (and dialysis machine), while protecting patient specific data (utilizing traceable security keys and signature metadata) and maintaining safe operating limits. Implementations of the present disclosure can also manage and deploy fault tolerant SHA-256 based “multimode encoded security keys” (e.g., traceable keys) that facilitate remote software updates, remote configuration management, and remote secure data transfer and management.

Implementations of the present disclosure can provide one or more of the following functionalities. Infection growths near patient access points can be monitored and prevented. Clots from stenosis can be monitored and detected. Pre-emptive interventions to fix an AVF can be facilitated. Continuous feedback can be provided to patients for self-management, safety triggered “call back message” (for example, when potassium levels exceed safe limits, blood pressure changes exceed the safe limits, etc.), which can provide feedback for the clinic/physician when the patient is away from the dialysis machine. Continuous kidney sensory related data can be provided to a physician during or after dialysis treatments. Furthermore, in some implementations, when a secure AI enabled wearable sensor platform is within the proximity of a security engine platform (which can be installed in dialysis machine), it can receive, after authentication and handshake protocols, raw sensor data (while away from dialysis and in-dialysis), processed sensor data for suggestions to the physician (for example, indicating key characteristics, trends, etc.), sensor system signature metadata (for example, per request from the Security Engine Platform), or a combination of them, among others.

In the following description, for the purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the present disclosure. It will be apparent, however, that the present disclosure may be practiced without these specific details. In other instances, well-known structures and devices are shown in block diagram form in order to avoid unnecessarily obscuring the present disclosure.

In the drawings, specific arrangements or orderings of schematic elements, such as those representing devices, modules, instruction blocks, and data elements, are shown for ease of description. However, it should be understood by those skilled in the art that the specific ordering or arrangement of the schematic elements in the drawings is not meant to imply that a particular order or sequence of processing, or separation of processes, is required. Further, the inclusion of a schematic element in a drawing is not meant to imply that such element is required in all embodiments or that the features represented by such element may not be included in or combined with other elements in some embodiments.

Further, in the drawings, where connecting elements, such as solid or dashed lines or arrows, are used to illustrate a connection, relationship, or association between or among two or more other schematic elements, the absence of any such connecting elements is not meant to imply that no connection, relationship, or association can exist. In other words, some connections, relationships, or associations between elements are not shown in the drawings so as not to obscure the disclosure. In addition, for ease of illustration, a single connecting element is used to represent multiple connections, relationships, or associations between elements. For example, where a connecting element represents a communication of signals, data, or instructions, it should be understood by those skilled in the art that such element represents one or multiple signal paths (e.g., a bus), as may be needed, to affect the communication.

Reference will now be made in detail to embodiments, examples of which are illustrated in the accompanying drawings. In the following detailed description, numerous specific details are set forth in order to provide a thorough understanding of the various described embodiments. However, it will be apparent to one of ordinary skill in the art that the various described embodiments may be practiced without these specific details. In other instances, well-known methods, procedures, components, circuits, and networks have not been described in detail so as not to unnecessarily obscure aspects of the embodiments.

Several features are described hereafter that can each be used independently of one another or with any combination of other features. However, any individual feature may not provide any of the advantages discussed above or may provide one of the advantages discussed above. Although headings are provided, data related to a particular heading, but not found in the section having that heading, may also be found elsewhere in this description.

FIG. 1 shows a perspective view of a hemodialysis system 100 that includes a hemodialysis machine 102 connected to a disposable blood component set 104 that partially forms a blood circuit. During hemodialysis treatment, an operator connects arterial and venous patient lines 106, 108 of the blood component set 104 to a patient. The blood component set 104 includes an air release device 112, which contains a self-sealing vent assembly that allows air but does not allow liquid to pass. As a result, if blood passing through the blood circuit during treatment contains air, the air release device 112 will vent the air to atmosphere.

The blood component set 104 is secured to a module 130 attached to the front of the hemodialysis machine 102. The module 130 includes the blood pump 132 capable of circulating blood through the blood circuit. The module 130 also includes various other instruments capable of monitoring the blood flowing through the blood circuit. The module 130 includes a door that when closed, as shown in FIG. 1, cooperates with the front face of the module 130 to form a compartment sized and shaped to receive the blood component set 104. In the closed position, the door presses certain blood components of the blood component set 104 against corresponding instruments exposed on the front face of the module 130.

The operator uses a blood pump module 134 to operate the blood pump 132. The blood pump module 134 may include a display window, a start/stop key, an up key, a down key, a level adjust key, and an arterial pressure port. The display window displays the blood flow rate setting during blood pump operation. The start/stop key starts and stops the blood pump 132. The up and down keys increase and decrease the speed of the blood pump 132. The level adjust key raises a level of fluid in an arterial drip chamber. The controls may exist as physical buttons/switches/actuators/etc. or wholly within a graphical user interface on a touch screen accessible display 118 or some combination thereof.

The hemodialysis machine 102 further includes a dialysate circuit formed by the dialyzer 110, various other dialysate components, and dialysate lines connected to the hemodialysis machine 102. Many of these dialysate components and dialysate lines are inside the housing 103 of the hemodialysis machine 102 and are thus not visible in FIG. 1. During treatment, while the blood pump 132 circulates blood through the blood circuit, dialysate pumps (not shown) circulate dialysate through the dialysate circuit.

A dialysate container 124 is connected to the hemodialysis machine 102 via a dialysate supply line 126. A drain line 128 and an ultrafiltration line 129 also extend from the hemodialysis machine 102. The dialysate supply line 126, the drain line 128, and the ultrafiltration line 129 are fluidly connected to the various dialysate components and dialysate lines inside the housing 103 of the hemodialysis machine 102 that form part of the dialysate circuit. During hemodialysis, the dialysate supply line 126 carries fresh dialysate from the dialysate container 124 to the portion of the dialysate circuit located inside the hemodialysis machine 102. As noted above, the fresh dialysate is circulated through various dialysate lines and dialysate components, including the dialyzer 110, that form the dialysate circuit. As will be described later, as the dialysate passes through the dialyzer 110, it collects toxins from the patient's blood. The resulting spent dialysate is carried from the dialysate circuit to a drain via the drain line 128. When ultrafiltration is performed during treatment, a combination of spent dialysate (described below) and excess fluid drawn from the patient is carried to the drain via the ultrafiltration line 129.

The dialyzer 110 serves as a filter for the patient's blood. The dialysate passes through the dialyzer 110 along with the blood, as described above. A semi-permeable structure (e.g., a semi-permeable membrane and/or semi-permeable microtubes) within the dialyzer 110 separates blood and dialysate passing through the dialyzer 110. This arrangement allows the dialysate to collect toxins from the patient's blood. The filtered blood exiting the dialyzer 110 is returned to the patient. The dialysate exiting the dialyzer 110 includes toxins removed from the blood and is commonly referred to as “spent dialysate.” The spent dialysate is routed from the dialyzer 110 to a drain.

A drug pump 192 also extends from the front of the hemodialysis machine 102. The drug pump 192 is a syringe pump that includes a clamping mechanism configured to retain a syringe 178 of the blood component set 104. The drug pump 192 also includes a stepper motor configured to move the plunger of the syringe 178 along the axis of the syringe 178. A shaft of the stepper motor is secured to the plunger in a manner such that when the stepper motor is operated in a first direction, the shaft forces the plunger into the syringe, and when operated in a second direction, the shaft pulls the plunger out of the syringe 178. The drug pump 192 can thus be used to inject a liquid drug (e.g., heparin) from the syringe 178 into the blood circuit via a drug delivery line 174 during use, or to draw liquid from the blood circuit into the syringe 178 via the drug delivery line 174 during use.

The hemodialysis machine 102 includes a security engine platform 105-1 that includes a control unit 101 (e.g., a processor) configured to receive signals from and transmit signals to the display 118, the control panel 120, and a communication module 107 (e.g., a near field communication (NFC) transceiver). The control unit 101 can also communicate with a network, server (e.g., an Internet server), or a master design and test platform (for example, as described later in this specification). In some implementations, the control unit 101 controls operating parameters of the hemodialysis machine 102 based at least in part on signals received by the display 118, the control panel 120, and the communication module 107.

The display 118 presents a user interface that may include vital signs of a patient, operational parameters of the dialysis treatment, and controls associated with the hemodialysis process. For example, the operational parameters may include ultrafiltration parameters, blood pump rate and information associated with the dialysate, hematocrit alert levels, blood pressure alarm limits, medicine infusion parameters, etc. The display 118 may include a touch screen through which the operator may interact and control the hemodialysis machine 102. For example, the operator may input various treatment parameters associated with the hemodialysis process.

The communication module 107 is configured to detect and communicate with an authenticated short-range wireless device—for example, an ID card 105, smart card, mobile device (for example, a smartphone belonging to a patient of medical care professional), and/or one of the wearable secure AI enabled sensor platforms described in this specification—when the device is within its wireless communication range. The communication module 107 is capable of communicating (for example, through multimode encoded security keys) information associated with the patient to the control unit 101. The communication module 107 can send a signal to the control unit 101 indicating that the patient is no longer present.

The control unit 101 is capable of receiving information (for example, through multimode encoded keys) associated with the identity of the patient from the communication module 107 and comparing the information with authentication profiles for multiple users (for example, patients) stored in a database. The database may be stored on a storage device associated with the dialysis system 100 or located at an external storage device (for example a server or a storage device associated with a different dialysis system.) Based on the comparison, the control unit 101 may retrieve the interface configuration profile of the operator or patient. The specific user interface that appears on the display may be based on the retrieved interface configuration profile. For example, the operating parameters and the dialysis controls that constitute the user interface may be determined from the retrieved interface configuration profile. Additionally, identity of the operator or patient, and the identity of a supervisor of the operator may also be displayed.

The control unit 101 is capable of communicating information associated with identity of the patient and a secure AI enabled wearable sensor platform (such as one or more of the wearable sensor platforms described later in this specification). The control unit 101, with a command and control instruction (for example, received from a master design and test platform described in this specification) is capable of adding or modifying an interface and/or operation configuration profile. The control unit 101 may be configured to detect an alert or alarm (used interchangeably herein) in the hemodialysis system 100, and based on the alarm, change the user interface that appears on the display 118 (for example, an alarm specific to a security and/or system anomaly). The alarm may be caused when the value of an operational parameter satisfies a criteria. For example, an alarm may occur when conductivity of the dialysate is above a certain value. This may require immediate attention from a nurse, and therefore the user interface that appears on the display 118 may be changed to a form expected by and/or familiar to the nurse.

FIG. 2 shows a perspective view of a peritoneal dialysis (PD) system 200 that includes a PD cycler (also referred to as a PD machine) 202 seated on a cart 204. The PD cycler 202 includes a housing 206, a door 208, and a cassette interface (not shown) that contacts a disposable PD cassette 212 when the cassette 212 is disposed within a cassette compartment formed between the cassette interface and the closed door 208. A heater tray 216 is positioned on top of the housing 206. The heater tray 216 is sized and shaped to accommodate a bag of dialysate (e.g., a 5 liter bag of dialysate). The PD cycler 202 also includes a display 118 and additional control panel 220 that can be operated by a user (e.g., a patient) to allow, for example, set-up, initiation, and/or termination of a PD treatment.

Dialysate bags 222 are suspended from fingers on the sides of the cart 204, and a heater bag 224 is positioned in the heater tray 216. The dialysate bags 222 and the heater bag 224 are connected to the cassette 212 via dialysate bag lines 226 and a heater bag line 228, respectively. The dialysate bag lines 226 may be used to pass dialysate from dialysate bags 222 to the cassette 212 during use, and the heater bag line 228 may be used to pass dialysate back and forth between the cassette 212 and the heater bag 224 during use. In addition, a patient line 230 and a drain line 232 are connected to the cassette 212. The patient line 230 may be connected to a patient's abdomen via a catheter and may be used to pass dialysate back and forth between the cassette 212 and the patient's peritoneal cavity during use. The drain line 232 may be connected to a drain or drain receptacle and may be used to pass dialysate from the cassette 212 to the drain or drain receptacle during use. The various components that pass dialysate form a dialysate circuit.

The PD system 200, like the hemodialysis system 100, includes a control unit 201, a display 218 and a communication module 207 that communicate with one another, and with a short-range wireless device (for example, the ID card 105 or a smartphone). Similar to the hemodialysis system 100, the PD system 200, using components such as the control unit 201 and the communication module 207, may detect information associated with the identity of an operator or patient, and change the user interface that appears on the display 218 based on an interface configuration profile. The PD system 200 allows an operator or patient to add or modify the interface configuration profile that may be stored in a storage device associated with the PD system 200, or an external storage device (for example, a server). Furthermore, the control unit 201 of the PD system 200 may change the user interface that appears on the display 218 based on an alarm in the PD system 200.

FIG. 3A shows a block diagram of a secure AI enabled wearable sensor platform 304, according to one or more implementations of the present disclosure. The wearable sensor platform 304 includes two or more sensors 312-1 to 312-N (hereinafter collectively referred to as the sensors 312). The sensors may take measurements in connection with a medical area, site or point of interest 306, for example, an access point for an AVF. For example, using the center of the medical point (e.g. access point) 306 for orientation, each of the sensors 312 is located a distance dx from the access point 306 at a direction ?x. The sensors 312 generate sensor data by measuring conditions at specific locations relative to the location of the medical point 306. It is noted that the system described herein is discussed principally in connection with medical sensor platform worn on the arm, e.g. in connection with monitoring a vascular access site; however, the wearable sensor platform of the system described herein may be appropriately utilized in connection with other medical areas and points of interest for a patient, and may be configured to be wearable in locations and manners other than on the arm.

The sensors 312 can include two or more of pressure sensors, temperature sensors, oxygen saturation SpO2 sensors, color sensors (RGB color or XYZ spectral sensors), infrared (IR) sensors, ultrasonic time-of-flight sensors, force sensors like accelerometers, among others. For example, the sensors 312 can include a pulse oximetry (SpO2) sensor that uses a red light source, an infrared light source, photo detectors, and a probe to measure oxygen saturation of a patient's blood. The sensors 312 can include a combination of an RGB Color sensors, XYZ spectral sensors, and infrared sensors that can facilitate determining one or more of: Blood CO2 saturation levels, potassium (K) concentration and quantity, Phosphate (PO3) concentration and quantity, possibility of Anemia, possibility of Fistula Infection (e.g., Fistula health, transfer efficiency, and so forth), and other health parameters (e.g., hormone levels).

The sensors 312 can include MEMS ultrasonic time-of-flight sensors to facilitate determining one or more of: blood flow rate, blood vessel health (e.g., determine degradation amounts for veins, arteries, capillaries, and so forth), and blood vessel transfer efficiency. The sensors 312 can include temperature sensors to facilitate determining one or more of the following: correlations between infection and a healthy access point and patient body responses during and after dialysis. The sensors 312 can include MEMS pressure sensors to facilitate determining one or more of: heart beat rate, blood pressure, correlations between blood flow rate and blood pressure (e.g., in conjunction with MEMS ultrasonic time-of-flight sensors), and heart condition (e.g., in conjunction with MEMS ultrasonic time-of-flight sensors).

In some implementations, two or more of the sensors 312 are tunable/adaptable to measure a desired characteristic, in which the two or more tunable/adaptable sensors 312 are tuned to specific wavelengths appropriate for measuring the desired characteristic of a particular patient. For example, at least some of the sensors 312 and light sources can be tuned to wavelengths appropriate for measuring the potassium levels of a particular patient's blood based on information learned about the particular patient, and then be tuned to wavelengths appropriate for measuring the phosphorous levels of the particular patient's blood based on information learned about the particular patient. The tuning/adapting of sensors to a particular patient is described later in this specification.

The wearable sensor platform 304 includes an analog front end (AFE) 314 for conditioning and processing the sensor data generated. Signal conditioning by the analog front end 314 can include preparing signal levels of sensor data of each of sensors 312 so that analogous data can be compared. Signal conditioning can include applying a Nyquist filter to the sensor data. Signal processing of the sensor data by the analog front end 314 can include using tunable hardware filters capable of extracting stable data of interest with minimal noise. Signal processing can include signal-noise-ratio optimization. For example, RGB IR optical sensors can be packaged with an RGB IR light source. During design of the sensors (for example, during a algorithm and hardware design phase at a master design and test platform as described later in this specification), a testing process can be used that simulates a patient's skin surface. For each target of interest (for example, sodium, potassium, SPO2, etc.), there can be a corresponding spectral frequency response. For a spectral response, signal to noise ratio (SNR) may depend on gain, bandwidth, nature of scattering (reflecting) surface, movements (skin) of the reflecting surface, intensity of light source, ambient light, distance from the reflecting medium, electronic noise (from circuitry), ground noise from power source, surrounding electromagnetic noise induction, and so forth. Based on these factors, outputs of the sensors can be baselined.

Signal processing can include applying one or more tracking algorithms to the conditioned sensor data. The tracking algorithms can be adapted to a particular characteristic of interest. For example, a spectrometer sensor can be tuned to detect potassium based on baseline data and dynamic data collected from the patient through the wearable sensor platform 304. Tuning and tracking algorithms can be defined specific to a sensor and a measured objective (which is independent of patient specific data). Baseline reference data can be created using a master design and test platform by inducing several (or all) possible real-life scenarios/variables. For example, a spectral sensor utilizing absorption/scattering property can be burdened with noise artifacts from skin color, skin movements, and so forth. AI-machine learning systems at the master design and test platform can have tunable algorithms and can use ideal spectral response data as baseline input, and then determine all possible conditions that can affect SNR. AI-machine learning algorithms can be used to create a recognizable higher confidence level to track and measure a desired measurement and/or characteristics of interest (signature metadata). The baseline algorithm and data can create a reference profile for the measurement of interest. When the wearable senor platform 304 is on a patient, the sensor data can be received by an artificial intelligence (AI) machine learning system 340 of the wearable sensor platform 304, and the AI machine learning system 340 can create a signature metadata. The AI machine learning system 340 can be used to compare the sensor data with the baseline data and provide appropriate control parameters to control the analog front end 314 to get the sensor data that can be close to the baseline data (signature/profile) created using the Master Design and Test Platform. This continuous real-time learning and continuous tuning to get optimized data (with optimized SNR) can provide predictive/deterministic measurements, increasing processing time while reducing power consumption (for the machine learning process).

Signal processing can also include one or more of applying sensor response/pattern reorganization algorithms to the conditioned sensor data and applying predictive algorithms to the conditioned sensor data. The analog front end 314 can include tunable filters that facilitate spectral response processing that can be adapted to a particular patient. The analog front end 314 can track the sensors and calibrate the sensors based on data obtained during tracking. The analog front end 314 can algorithmically control the sensors 312 to tune one or more of the sensors to a particular patient such that desired characteristics of the particular patient can be measured, as described previously (e.g., tune an RGB sensor to a particular range of wavelengths so that the sensor data can be used to measure a potassium concentration of the patient). In some implementations, the analog front end 314 applies sensor response/pattern reorganization and predictive algorithms to the sensor data. As will be explained later in this specification, the algorithms can be designed such that they do not infringe upon specified rules and boundary conditions.

The wearable sensor platform 304 includes a processor 316 which performs further preparation of the sensor data for transmission. The processor 316 can be an ultra-low power processor, a field programmable gate array (FPGA), a complex programmable logic device (CPLD), and so on.

The secure AI enabled wearable sensor platform 304 (sometimes referred to as a wearable sensor platform 304 in this specification) includes a wireless communications interface 318 (sometimes referred to as a communications interface 318 or interface 318 in this specification). The communications interface 318 allows communication with one or more external devices, such as a security engine platform residing on a dialysis system (as described previously with respect to FIGS. 1-2. The wearable sensor platform 304, after a successful authentication with a security engine platform, can send status and monitoring data prepared by the security engine platform. The wearable sensor platform 304 can also receive configuration updates, software updates, and calibration updates from the security engine platform. The interface 318 can includes radio frequency transceivers. Examples of interfaces 318 include radio transceivers compatible with several Wi-Fi standards, cellular modems, and so forth.

The wearable sensor platform 304 includes a display 320. The display 320 can include one or more light emitting diode (LED) lights for providing visual indications. The LED lights can indicate that the wearable sensor platform 304 is functioning/not functioning properly and/or the access point 306 is/isn't at risk. In some embodiments, since functioning properly is expected most of the time, the LED lights only turn on to indicate that something is wrong; this way, the LED lights do not consume energy during normal operation. The display 320 can also include one or more auditory indicators like a speaker that beeps when the wearable sensor platform 304 is either not functioning properly or when the access point 306 is at risk. In some embodiments, the display 320 provides a visual and/or auditory indication when the wearable sensor platform 304 first comes online, when the wearable sensor platform 304 is proximate to the dialysis machine (e.g., when the wearable sensor platform 304 detects the presence of the security engine platform by handshake protocols), and when blood flow rate sensor data profiles match to flow rate data during a dialysis treatment.

The wearable sensor platform 304 includes a power source 322 for providing power to the different components described. In some embodiments, the power source 322 is a battery or a rechargeable battery.

The wearable sensor platform 304 includes a security enforcer 323. Generally, the security enforcer 323 is capable of facilitating the updating of the AI machine learning system 340 using a key-based authentication scheme, guard-banding operational/control limits governed by the AI machine learning system 340, providing data (e.g., raw sensor data and process data) to a physician (e.g., via a security engine platform residing in a dialysis system) to assist the physician in deriving quicker and effective decisions regarding a patient and treatments (e.g., prescriptions). The security enforces is also generally capable of autonomously facilitating warning/messaging decisions when any of the data measured by the sensors 312 triggers a critical safety state (e.g., blood pressure reading exceeds safe low and high limits).

The security enforcer 323 is programmable hardware, e.g., an FPGA or a CPLD that serves one or more functions. In some implementations, the security enforcer 323 performs platform management, power management, and security enforcement. For platform management, the security enforcer 323 manages system level critical tasks for the wearable sensor platform 304. Platform management can include ensuring sensor data integrity and operational stability control. For example, the security enforcer 323 receives data from the sensor 312-1. The data from the sensor 312-1 is appended with an authentication code where the security enforcer 323 checks this authentication code against a code from the sensor 312-1 to determine whether the data came from the specific sensor.

Platform management can also include monitoring safety critical status signals and critical signal integrity. Safety critical status signals are signals assigned to represent safe operating limits of the wearable sensor platform 304. An example includes power monitoring signals that define status of the power source 322. The power monitoring signals can indicate a remaining power level where the security enforcer 323 determines whether the remaining power level is adequate to provide trustable information from the sensors 312. For example, if remaining power level falls below a predefined level, then the security enforcer 323 enforces safety protocols. The safety protocols include operational safety protocols and/or human safety protocols. An operational safety protocol can involve storing present data, setting a warning indicator (e.g., blinking orange light from a tri-color LED of the status indicator 320), reducing sampling rate for the sensors 312, and reducing signal processing rate at the analog front end 314 and/or the processor 316. In some implementations, the security enforcer 323 performs human operational control, which can include sounding an alarm via speakers in the display 320 to check for a short circuit that can create high current and may cause burning.

In some embodiments, signal integrity of safety critical status signals cannot be trusted to enforce correct safety protocols. The security enforcer 323 determines whether the signal has too much noise susceptibility. Signals with too much noise susceptibility include, for example, unstable signals due to susceptibility to changes in ambient temperature and unstable signals due to electromagnetic compatibility (EMC) compliances. The security enforcer 323 also monitors abnormal loading of configuration and calibration data. In some implementations, authentication of the security enforcer 323 is required to modify the configuration and calibration data.

Platform management can also include emergency control enforcing for safety critical parameters and functionality. Platform management can also include sensor calibration status check with respect to stored baseline data. Platform management can also include control and retention of factory calibration data, sensor functional/operational configuration data, and/or the wearable sensor platform 304 functional/operational configuration data. Platform management can further include system alarm and safe operating limits management. Platform management can also include safeguarding boot code of the security enforcer 323 from corruption from one or more forms of unauthorized/undesired changes.

Power management of the security enforcer 323 involves power sequencing and reset processing, power monitoring, power management and optimizations for longer time-between-recharge and for performance optimizations, and power emergency controls. Power emergency controls include safety protocol management and safe management of sensor data when available operating power is below a safe operating threshold, as discussed above.

Security enforcement of the security enforcer 323 includes cybersecurity guard/compliance enforcement for secure and safe operation of hardware and software of the wearable sensor platform 304. Security enforcement includes security key management for master authentication and encryption and decryption engines for secure signal and/or data processing, transfer, and storage. In master authentication, a security key is transferred from a manufacturing site of the wearable sensor platform 304 (after factory acceptance tests) to authenticate and enable secure engine functionality. At the manufacturing site, device operational configuration files can be loaded, calibrated, and tested before approval for customer delivery. Security enforcement also includes threat monitoring and critical event logging, e.g., logging of safety and security events, logging of remote accesses to the wearable sensor platform, and so on. Security enforcement can also include a safety and security mode (including internal and external data management and communication/data transfer).

In some implementations, security enforcement includes receiving a multi-mode encoded key (e.g., traceable key) from an outside source (e.g., a standalone board residing in a dialysis machine). In some implementations, the multi-mode encoded key is designed by the manufacturing site (for example, a master design and test platform as described later in this specification). The multi-mode encoded key can correspond with a secure hash algorithm (SHA), such as an SHA256 algorithm. The security enforcer 323 can process the multimode encoded keys to generate an identification key particular to the patient using the wearable sensor platform 304, and use the identification key to encrypt the raw sensor data captured by the sensors 312, such that the encrypted raw sensor data is traceable to the particular patient. As will be described later, the encrypted raw sensor data can be transmitted to an outside source (e.g., a security engine platform as described in this specification) for further processing. The security enforcer 323 can also receive the multimode encoded keys from a security engine platform to authenticate the wearable sensor platform 304 to receive software updates from the security engine platform (e.g., updates to machine learning algorithms, rules and boundary conditions, and so forth, as explained later in more detail).

Security enforcement by the security enforcer 323 also includes supervising the AI machine learning system 340. For example, the security enforcer 323 can guard-band operational/control limits governed by the AI machine learning system 340, as described later in more detail. The guard-banding can, for example, ensure that algorithmic modifications to the sensors 312 and/or the analog front end 314 do not cause these components to operate outside of their pre-validated state (e.g., in accordance with validation performed by a manufacturer/Master Design and Test Platform, as described later in more detail).

The AI machine learning system 340 applies rules/goals based (e.g., baseline rules/goals that are designed and validated by a mater design and test platform) machine learning techniques to train the machine learning model 340a that, when applied to input data, generates indications of whether the input data items have an associated property or properties. As part of the training of the machine learning model 340a, the AI machine learning system 340 can form a training set of input data by identifying a positive training set of input data items that have been determined to have the property in question, and, in some implementations, forms a negative training set of input data items that lack the property in question.

The AI machine learning system 340 extracts values from the input data of the training set, the values being variables deemed potentially relevant to whether or not the input data items have the associated property or properties (e.g., defined by the rules/goals). An ordered list of the features for the input data is herein referred to as the feature vector for the input data. In some implementations, the AI machine learning system 340 applies dimensionality reduction (e.g., via linear discriminant analysis (LDA), principle component analysis (PCA), learned deep features from a neural network, or the like) to reduce the amount of data in the feature vectors for the input data to a smaller, more representative set of data.

In some implementations, the AI machine learning system 340 uses supervised machine learning (for example, at a master design and test platform to create baseline profile data for each sensors and corresponding measurements) to train the machine learning model 340a with the feature vectors of the positive training set and the negative training set serving as the inputs. Different machine learning techniques—such as linear support vector machine (linear SVM), boosting for other algorithms (e.g., AdaBoost), neural networks, logistic regression, naïve Bayes, memory-based learning, random forests, bagged trees, decision trees, boosted trees, or boosted stumps—are used in some implementations. The machine learning model 340a, when applied to the feature vector extracted from the input data item, outputs an indication of whether the input data item has the property in question, such as a Boolean yes/no estimate, a scalar value representing a probability, a vector of scalar values representing multiple properties, or a nonparametric distribution of scalar values representing different and not a priori fixed numbers of multiple properties, which may be represented either explicitly or implicitly in a Hilbert or similar infinite dimensional space.

In some implementations, a validation set is formed of additional input data, other than those in the training sets, which have already been determined to have or to lack the property in question. The AI machine learning system 340 applies the trained machine learning model 340a to the data of the validation set to quantify the accuracy of the machine learning model 340a. Common metrics applied in accuracy measurement include: Precision=TP/(TP+FP) and Recall=TP/(TP+FN), where precision is how many the machine learning model 340a correctly predicted (TP or true positives) out of the total it predicted (TP+FP or false positives), and recall is how many the machine learning model 340a correctly predicted (TP) out of the total number of input data items that did have the property in question (TP+FN or false negatives). The F score (F-score=2*PR/(P+R)) unifies precision and recall into a single measure. In some implementations, the machine learning system 340 iteratively re-trains the machine learning model 340a until the occurrence of a stopping condition, such as the accuracy measurement indication that the model 340a is sufficiently accurate, or a number of training rounds having taken place.

In some implementations, the machine learning model 340a includes a neural network. In some implementations, the neural network includes a convolutional neural network. The machine learning model 340a can include other types of neural networks, such as recurrent neural networks, radial basis function neural networks, physical neural networks (e.g., optical neural network), and so forth.

The machine learning model 340a is configured to modify, based on learned features particular to the patient using the wearable sensor platform 304, the algorithms used by the analog front end 314 to perform signal processing. For example, based on learned characteristics of the patient, the machine learning model 340a can tune the tunable filters of the analog front end 314 to a particular patient to facilitate adaptable spectral response processing. The machine learning model 340a can also update the user adaptable tracking algorithms and SNR optimization algorithms of the analog front end 314, based on learned characteristics of the patient. The machine learning model 340a can implement rules and boundary conditions set forth by the security enforcer 323 to ensure that any algorithmic updates do not cause the wearable sensor platform 304 to infringe validated states as set forth by a manufacturer/Master Design and Test Platform (described later). The machine learning model 340a can include a safe-operation look-up table that consists of operating parameters and limits related to basic safety and essential performances. The safe-operation look-up table can be used to create rules and constraints for the AI-machine learning system to protect against any system changes that will induce non-validated states or conditions for the platform 304. The machine learning model 340a can use function enable keys to facilitate changes or modifications, or enable/disable functionality with respect to the AI-machine learning system.

FIG. 3B shows a block diagram of an environment 300 for monitoring a vascular access, according to one or more implementations of the present disclosure. The environment includes and Master Design and Test (MDT) platform 370 (which can be integrated with a security engine platform 361), a dialysis system 360, and the wearable sensor platform 304 described previously with reference to FIG. 3A.

In general, the MDT platform 370 provides remote software updates and remote diagnostics for the wearable sensor platform 304 and the security engine platform 361 of the dialysis system 360. The MDT platform 370 is also generally capable of providing deep root-cause analysis for the security engine 361 of the dialysis system 360 and wearable sensor platform 304 (e.g., any root-cause analysis that is beyond the capabilities of the security engine platform 361 and wearable sensor platform 304). The MDT platform 370 is generally capable of providing the backbone for deep trend analysis and comparative (snapshot) studies among several security engines 361, along with continuous design improvements, rule/goal tables, and so forth.

The MDT platform 370 is configured to generate and validate the multi-mode encoded keys described previously with reference to FIG. 3A. The MDT platform 370 is configured to provide model based design and development environments for the design of hardware, software, and algorithms used with the wearable sensor platform 304 and the security engine platform 361 of the dialysis system 360, as well as the optimization, verification, and validation of such algorithms.

The dialysis system 360 can be similar to the dialysis systems 100, 200 described previously with reference to FIGS. 1-2. The dialysis system 360 includes the security engine platform 361, which can be a stand-alone board that resides in dialysis equipment of the dialysis system 360. The security engine platform 361 can authenticate the wearable sensor platform 304 and its functions. The security engine 361 can receive the multi-mode encoded keys from the MDT platform 370, and provide the keys to the wearable sensor platform 304 (e.g., when the wearable sensor platform 304 is within a close range due to the patient undergoing dialysis treatment). The security engine platform 361 can use the multi-mode encoded keys to authenticate and encrypt/decrypt data transferred between the security enforcer 323 of the wearable sensor platform 304 and the security engine 361. The security engine platform 361 is configured to perform software updates and lifecycle management of the wearable sensor platform 304. The security engine platform 361 is configured to receive one or more of: the captured raw sensor data from the wearable sensor platform 304 and the processed sensor data from the wearable sensor platform 304. As indicated previously, the captured raw sensor data and/or the processed sensor data can be encrypted by the wearable sensor platform 304 (e.g., using the security enforcer 323) with a multi-mode encoded keys. In some implementations, the security engine platform 361 can include corresponding multi-mode encoded keys which can be used to decrypt the received data. The security engine 361 can transmit the received data to the MDT platform 370 before or after encryption. In some implementations, the MDT platform 370 includes corresponding multi-mode encoded keys for decrypting the data. As indicated previously, the processed sensor data can include intelligent information. The security engine 361 can provide the raw captured data and the processed sensor data, along with encrypted prescription data representing prescriptions used by the dialysis system 360 to perform treatment and performance data indicating how the dialysis system 360 performed for a specific prescription, to a physician (e.g., to an electronic device of the physician or through a user interface of the dialysis machine).

FIG. 4 shows a flowchart of a method 400 for monitoring the health of a patient, according to one or more implementations of the present disclosure. For illustrative purposes, the method 400 is described as being performed by the wearable sensor platform 304, as described previously with reference to FIG. 3A. The method 400 includes one or more of: receiving physiological measurements (block 402), determining differential readings (block 404); determining if the differential readings exceed an alarm threshold (block 405); determining if physiological measurements approximately equal a nominal value (block 406); determining if the physiological measurements deviate from the nominal value by a threshold value (block 407); determining trend information (block 408); determining whether to generate an alert based on the trend information (block 410); and generating an alert (block 409), which may include displaying the alert on an appropriate display, as further discussed herein.

At block 402, data representing physiological measurements captured at different sensor locations of the wearable sensor platform 304 are received by the analog front end 314.

At block 404, the analog front end 314 provides normalized raw data to the AI machine learning system 340 to determine differential readings between the different sensor locations based on the physiological measurements. The analog front end incorporates mutual distances between the sensor locations and distances between the access point 306 and the sensor locations with the physiological measurements to determine the differential readings between the different sensor locations.

At block 405, in some implementations, the analog front end 314, in conjunction with the AI machine learning system 340, determines whether the differential readings exceed an alarm threshold. The alarm threshold can be determined by the machine learning model 340a based on information learned about the user of the wearable sensor platform 304. For example, the machine learning model 340a can learn a standard baseline reference (sometimes referred to as a static threshold in this specification) temperature near a fistula (for example, from a plurality of patients) and identify a new threshold for a particular patient based on sensor data collected over time by the wearable sensor platform 304 (sometimes referred to as a dynamic threshold in this specification). The machine learning model 340a can then use both the static threshold and the dynamic threshold to identify undesired states near the fistula. If, for example, an alarm threshold is exceeded, then at block 409, the security enforcer 323 generates an alarm. Generating an alarm can include causing the display 320 to exhibit visual cues (e.g., blinking lights) and/or acoustical cues (e.g., beeping sounds). Generating an alarm can include alerting an external systems/devices, such as the dialysis system 360, a mobile device of a physician, or a mobile device of the user of the wearable sensor platform 304.

At block 406, the analog front end 314 determines whether the physiological measurements approximate a nominal value. In some implementations, the physiological measurements approximate the nominal value when the physiological measurements are within a deviation percentage of 1% of the nominal value. In some implementations, the nominal value and the deviation percentage is determined by the machine learning model 340a based on information learned about the patient. A state where the physiological measurements approximate the nominal value may be referred to as an equilibrium state.

At block 407, the analog front end 314 determines whether the physiological measurements deviate from the nominal value by a threshold. The threshold can be determined by the machine learning model 340a based on information learned about the patient. In some implementations, if the analog front end 314 and the AI machine learning system 340 determine that the physiological measurements deviate from the nominal value by a threshold, then at block 309, the security enforcer 323 generates an alarm, as discussed previously.

In some implementations, at block 408, if the physiological measurements deviate from the nominal value by the threshold, the analog front end 314 and the AI machine learning system 340 determine trend information of the physiological measurements. Trend information includes tracking the physiological measurements for a specified time period to determine whether the physiological measurements are steady, are rising, or are falling. A state where trend information is determined can be called a caution state. The specified time period can be determined by the machine learning model 340a based on information learned about the user. In some implementations, the on/off sampling rate of a sensor is based on rates of change (and trend/variance results from the AI machine learning system 340) in data within a defined time window, power management rules and constraints (e.g., due to the wearable nature of the platform 304).

At 410, the analog front end 314 determines whether to generate an alert based on the trend information. For example, if the trend information shows certain measurements rising faster than a threshold rate of change, then the security enforcer 323 can generate an alert. The threshold rates of change can be user specific and determined by the machine learning model 340a based on information learned about the user.

As illustrated in FIG. 4, in some implementations, the analog front end 314 can receive physiological measurements at block 402 and then proceed to block 406 to determine whether the physiological measurements approximately equate the nominal value.

FIG. 5 shows a secure AI enabled wearable sensor platform 500, implemented as a sleeve worn on an arm of a patient, having temperature sensors for differential temperature measurements, according to one or more implementations of the present disclosure. In some implementations, the wearable sensor platform 304 discussed previously with reference to FIG. 3A includes some or all aspects of the wearable sensor platform 500. The wearable sensor platform 500 includes an analog front end 514, a security enforcer 523, a processor 516, a machine learning system 540, and a display 520, which can be substantially similar to corresponding components of the wearable sensor platform 304 discussed previously with reference to FIG. 3A.

The wearable sensor platform 500 includes a first temperature sensor 504-1, a second temperature sensor 504-2, and a third temperature sensor 504-3, each placed at different locations from an access point 502 and communicatively coupled to the analog front end 514. In some implementations, the first temperature sensor 504-1 and the second temperature sensor 504-2 are placed at opposite directions from the access point 502. In the illustrated example, the access point 502 is located on one side of the arm, and the third temperature sensor 504-3 is located on the opposite side of the arm. The third temperature sensor 504-3 location is chosen to be as far away from the access point 502 but equidistant to the first and second sensors 504-1, 504-2.

The spatial distance from the access point 502 to the first temperature sensor 504-1 is d1. The spatial distance from the access point 502 to the second temperature sensor 504-2 is d2. The spatial distance from the first temperature sensor 504-1 to the third temperature sensor 504-3 is d3, and the spatial distance from the second temperature sensor 504-2 to the third temperature sensor 504-3 is also d3.

In some implementations, having three readings from three different locations provides an area equivalent calculation for determining qualities of the access point 502. The different readings at the different locations when combined allow for performing trend analysis, developing standard deviations, and developing temperature data profiles and tracking algorithms. The multiple sensor configuration can provide deterministic and highly-reliable system decisions based on triple-redundant sensor data from two or more sensors. This allows decision based voting where a majority of sensors agreeing can be used as a strong bias towards determining whether a minority of sensors are faulty. In some implementations, the temperature sensors 504-1, 504-2, 504-3 sample body temperature, in degrees Celsius, at their respective locations once every minute.

The wearable sensor platform 500 can be configured to perform the method 400 discussed previously with reference to FIG. 4. For example, at block 402, the temperature sensors 504-1, 504-2, 504-3 can collect sensor data at a predetermined sampling rate (e.g., every minute) and sends the sensor data after a prescribed amount of time to the analog front end 514. The sampling rate can be determined by the AI machine learning system 540 based on information learned about the user, and governed by the security enforcer 523 such that the sampling rate is more particular to the user but stays within a validated/safe state.

At block 404, using the spatial distances d1, d2, and d3, and the temperature measurements in the sensor data of block 402, the analog front end 514 can then determine differential temperature readings between the different sensors as provided in Eqs. 1-3. d1 and d2 can be 3-5 cm from the access point 502.



Eq. 1 is the differential temperature reading between the first temperature sensor 504-1 and the second temperature sensor 504-2, where T1 and T2 are temperature measurements from the first and second temperature sensors 504-1, 504-2, respectively. Eq. 2 is the differential temperature reading between the first and third temperature sensors 504-1, 504-3, where T3 is the temperature measurements from the third temperature sensor 504-3. Eq. 3 is the differential temperature reading between the second and third temperature sensors 504-2, 504-3. T1, T2 and T3 are temperature measurements over a time period (t.stop-t.start). Thus, Eqs. 1-3 find the difference between average temperature readings of the first and second temperature sensors 504-1, 504-2, the first and third temperature sensors 504-1, 504-3, and the second and third temperature sensors 504-2, 504-3, respectively.

From the temperature measurements T1, T2 and T3 and the differential temperature readings of Eqs. 1-3, at block 406, the analog front end 514 determines whether each temperature measurement approximates a nominal temperature Tnom. In some implementations, temperature measurements T1, T2 and T3 should be within a deviation percentage of 1% of the nominal temperature Tnom. However, the deviation percentage can be determined by the machine learning system 540 as described previously. In the illustrated implementations, when T1, T2 and T3 are within 1% of Tnom, the wearable sensor platform 500 maintains an equilibrium state. When any of T1, T2 or T3 is not within 1% of Tnom, the wearable sensor platform 500 transitions to a caution state. The caution state signifies that the body temperature is either too high or too low. As described previously, Tnom can be determined by the machine learning system 540 based on information learned about the user.

In some implementations, at block 405, the differential temperature readings of Eqs. 1-3 can be used to determine whether the differential temperature readings exceed an alarm threshold. As discussed previously, the alarm threshold Talarm, can be determined by the machine learning system 540 based on information learned about the user. In some implementations, when Eq. 1, 2 or 3 is nonzero and the magnitude of Eq. 1, 2 or 3 evaluated exceeds an alarm threshold Talarm, then an alarm is generated at block 409. Eq. 4 summarizes this condition.


In Eqs. 1-4, the following definitions apply:

Talarm is Tnom×1%
Tnom is normal body temperature computed within 24 hours and can be determined by the machine learning system 540 as described previously.
Measurements T1, T2 and T3 made with the temperature sensors provide values for body temperature Tbody that can be used to determine a body temperature trend bandwidth Tbody±dT.

At block 408, the client device determines trend information using the temperature measurements if the physiological measurements are determined at block 407 to deviate from Tnom by more than 1%. In the event that the wearable sensor platform 500 transitions to a caution state, the temperature sensors 504-1, 504-2, 504-3 will change their sampling rates of temperature measurements T1, T2 and T3 from a slower sampling rate to a faster sampling rate.SECURE ARTIFICIAL INTELLIGENCE ENABLED MEDICAL SENSOR PLATFORMS
A secure artificial intelligence (AI) enabled wearable medical sensor platform is used for adaptive operation according to features and techniques described herein. Operational parameters are modified based on data inputs thereto that provide feedback to the AI systems of the wearable sensor platform. The described technology can facilitate adaptive optimizations provided by AI machine learning algorithms in a manner that can beneficially assist in the monitoring and treatment of a patient. For example, the system described herein may be used for the continuous monitoring of the physiological parameters and health of a patient's vascular access point (for example, the fistula) and may provide, among other things, early warnings of possible infection at the vascular access location.


This application is a continuation application of and claims the benefit of priority to U.S. application Ser. No. 16/897,896, filed on Jun. 10, 2020, the contents of which are hereby incorporated by reference.

This disclosure generally relates to sensor platforms for monitoring the health of a user including secure artificial intelligence (AI) enabled wearable sensor platforms.

Patients with kidney failure or partial kidney failure typically undergo dialysis treatment using medical devices such as dialysis machines. The two principal dialysis methods are hemodialysis (HD) and peritoneal dialysis (PD). During hemodialysis, the patient's blood is passed through a dialyzer of a hemodialysis machine while also passing dialysate through the dialyzer. A semi-permeable membrane in the dialyzer separates the blood from the dialysate within the dialyzer and allows diffusion and osmosis exchanges to take place between the dialysate and the blood stream. Before the blood is returned to the body, air bubbles are removed from the blood to inhibit embolisms. The process of removing air is typically accomplished through use of a venous drip chamber, which is located downstream of the blood outlet of a dialyzer and upstream of the venous blood return of the patient.

During peritoneal dialysis, the patient's peritoneal cavity is periodically infused with dialysate, or dialysis solution. The membranous lining of the patient's peritoneum acts as a natural semi-permeable membrane that allows diffusion and osmosis exchanges to take place between the solution and the blood stream. Automated peritoneal dialysis machines, also called PD cyclers, are designed to control the entire peritoneal dialysis process so that it can be performed at home, usually overnight, without clinical staff in attendance. Both HD and PD machines may include displays with touch screens or other user interfaces that display information of a dialysis treatment and/or enable an operator or patient to interact with the machine.

When healthy, kidneys maintain the body's internal equilibrium of water and minerals (e.g., sodium, potassium, chloride, calcium, phosphorous, magnesium, and sulfate). In hemodialysis, blood is taken from a patient through an intake needle (or catheter) which draws blood from an artery located in a specific accepted access location (arm, thigh, subclavian, etc.). The drawn blood is pumped through extracorporeal tubing via a peristaltic pump, and then through a special filter called a dialyzer. The dialyzer is intended to remove unwanted toxins such as blood urea, nitrogen, potassium, and excess water from the blood. The dialyzed blood then flows out of the dialyzer via tubing and a needle (or catheter) back into the patient.

Vascular access is used for hemodialysis. These accesses are expected to sustain blood flow rates above 300 mL/min. One type of access is the arteriovenous fistula (AVF). AVF is considered to be the most effective of the vascular accesses for dialysis patients. One reason AVFs are effective is that they use native blood vessels, which, compared to synthetic grafts, are less likely to develop stenosis and fail. AVFs for hemodialysis typically include a surgically created connection between an artery and a vein in the arm. Artery walls are thicker than vein walls, and since the AVF connects the artery and the vein, the walls of the veins become thicker due to higher arterial pressure. The strengthened vein can then tolerate needles during hemodialysis.

Many different types of systems benefit from adaptive operation that is provided when used with artificial intelligence. In such systems, operational parameters are modified based on data inputs thereto that provide feedback to the artificial intelligence systems.

SUMMARY
In at least one aspect of the present disclosure a wearable sensor platform is provided. The wearable sensor platform includes: a plurality of sensors configured to capture raw sensor data; a transceiver; one or more storage devices storing computer-executable instructions; and one or more computer processors configured to execute the computer-executable instructions to cause the one or more computer processors to perform operations. The operations include: adjusting, using a machine learning model, one or more algorithms for processing the captured raw sensor data based on user data of a user of the wearable sensor platform; receiving the captured raw sensor data from the one or more sensors; determining, using the adjusted one or more algorithms and based on the captured raw sensor data, differential readings between pairs of sensors of the plurality of sensors; determining, using the adjusted one or more algorithms, whether the differential readings exceed a threshold; and generating an alert in response to determining that the differential readings exceed the threshold.

The operations can further include: receiving constraint data indicating one or more constraints corresponding to the wearable sensor platform; and applying the constraint data to the machine learning model during the adjusting. The operations can further include receiving, from an external source, one or more encryption keys; encrypting, use the one or more encryption keys, the captured raw sensor data; and transmitting the captured raw sensor data to the external source using the transceiver. The external source can include one or more processors of a dialysis system. The one or more encryption keys can include one or more multi-mode encoded keys. The threshold can be based on baseline reference data of the user learned by the machine learning model. The operations can further include: determining, using the adjusted one or more algorithms and based on the captured sensor data, whether one or more physiological measurements of the user deviate from a nominal value by a second threshold; and determining, in response to determining that one or more physiological measurements of the user deviated from the nominal value by the second threshold, trend information for the one or more physiological measurements, the trend information indicating a change in the one or more physiological measurements over time.

These and other aspects, features, and implementations can be expressed as methods, apparatus, systems, components, program products, methods of doing business, means or steps for performing a function, and in other ways, and will become apparent from the following descriptions, including the claims.

Implementations can include one or more of the following advantages. Continuous monitoring of the physiological parameters and health of a patient can be accomplished to provide, among other things, early warnings of possible problems. The described technology can facilitate adaptive optimizations provided by machine learning algorithms in a manner that can assist in providing a physiological profile tuned to a user's natural physical state while a sensor system retains validated baseline parameters and a baseline threshold derived from, for example, national standards. The described technology further provides for enhanced security related to the capture and use of a patient's data.

For example, the system described herein may be used for the continuous monitoring of the physiological parameters and health of a patient's vascular access point (for example, the fistula) and may provide, among other things, early warnings of possible infection at the vascular access location. Critical physical (corporal) characteristics near a patient's access point can be continuously monitored. The described technology can be used to increase cost saving in maintaining a vascular access point of the user. Further, the described technology can be used to increase the blood transfer efficiency through the vascular access point, and increase the lifespan and reduction in maintenance cost related to the vascular access point.

These and other aspects, features, and implementations will become apparent from the following descriptions, including the claims.

BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 shows a perspective view of a hemodialysis system, according to one or more implementations of the present disclosure.

FIG. 2 shows a perspective view of a peritoneal dialysis system, according to one or more implementations of the present disclosure.

FIG. 3A shows a block diagram of a secure AI enabled wearable sensor platform, according to one or more implementations of the present disclosure.

FIG. 3B shows a block diagram of an environment for monitoring a vascular access, according to one or more implementations of the present disclosure.

FIG. 4 shows a flowchart of a method for monitoring the health of a patient, according to one or more implementations of the present disclosure.

FIG. 5 shows a secure AI enabled wearable sensor platform having temperature sensors for differential temperature measurements, according to one or more implementations of the present disclosure.

FIG. 6 shows a secure AI enabled wearable sensor platform having temperature sensors for gradient temperature measurements, according to one or more implementations of the present disclosure.

FIG. 7 shows a secure AI enabled wearable sensor platform having pressure sensors, according to one or more implementations of the present disclosure.

FIG. 8 shows a secure AI enabled wearable sensor platform having spectral sensors, according to one or more implementations of the present disclosure.

FIG. 9 shows a block diagram of an environment including a factory module, a supervisor module, and a plurality of operations modules, according to one or more implementations of the present disclosure.

FIG. 10 shows a flowchart of a method performed in connection with transferring calibration and configuration data, baseline reference data, and security keys from a factory module to a supervisor module, according to one or more implementations of the present disclosure.

FIG. 11A shows a flowchart of a method performed by an AI-machine learning system in conjunction with a security enforcer, according to one or more implementations of the present disclosure.

FIG. 11B shows a flowchart of a method performed by a security enforcer in conjunction with an AI-machine learning system, according to one or more implementations of the present disclosure.

FIG. 12 shows a block diagram of an environment including an AI-machine learning system, a security engine, a dynamic operational data set, and a baseline reference data set, according to one or more implementations of the present disclosure.

FIG. 13 is a block diagram of an example computer system used to provide computational functionalities associated with described algorithms, methods, functions, processes, flows, and procedures described in the present disclosure.

FIG. 14 is a schematic diagram illustrating information flow and interaction between different components and modules of a wearable sensor platform.

FIG. 15 is a schematic diagram showing an example of a connected medical information distribution system, for example, a connected health (CH) system, that may that may be used in connection with network aspects of the wearable sensor platform described herein.

DETAILED DESCRIPTION
A secure AI enabled wearable medical sensor platform may be used for adaptive operation according to features and techniques described herein. Operational parameters are modified based on data inputs thereto that provide feedback to the AI systems of the wearable sensor platform. The described technology can facilitate adaptive optimizations provided by AI machine learning algorithms in a manner that can beneficially assist in the monitoring and treatment of a patient. For example, the system described herein may be used for the continuous monitoring of the physiological parameters and health of a patient's vascular access point (for example, the fistula) and may provide, among other things, early warnings of possible infection at the vascular access location.

Although AVFs are effective accesses for patients undergoing hemodialysis, there are constraints that lead AVFs to fail. AVFs can become infected because they are exposed sites that facilitate removing blood from patients and reintroducing blood to the patients. Patients undergo hemodialysis about three times a week so there are multiple opportunities during the week for the AVF to become infected. Furthermore, AVFs can fail due to stenosis, improper maintenance during maturation period, hematoma, and a patient's overall lack of awareness. Typically, current AVF monitoring technologies are not wearable and do not perform continuous monitoring functions. Also, current AVF monitoring technologies can be invasive and are usually performed once every four weeks (and/or while a patient is receiving dialysis treatment). In these technologies, blood is recirculated, and an ultrasound or ultrasonic sensor is used to monitor the blood recirculation to determine whether the AVF is performing well. These technologies may also need blood chemistry lab results to measure various chemical and biological parameters.

Furthermore, the effectiveness of dialysis can depend on the patient's chemical, biological, and physical characteristics. A relatively healthy kidney may include natural biological sensory that can facilitate the maintenance of a patient's corporeal wellbeing, such as blood pressure control and production of hemoglobin. However, in a patient with chronic kidney disease (CKD), many of these natural biological sensors can become ineffective, and the corresponding natural corporeal sensory feedback control mechanisms can become less effective. For example, the lack of effective natural sensors embedded in the kidney can introduce unpredictable and/or unmanageable changes in the state of natural health of the patient (e.g., deficiency in blood pressure control, deficiency in the rate of hemoglobin production, and so forth). Furthermore, in systems that implement AI-machine learning, operational parameters can be modified based on sensor data inputs that provide feedback to the machine learning systems. For example, a speech recognition system may adapt and improve based on data input that indicates successes and failures. The adaptive optimizations may be provided by artificial intelligence processing that revises operation states of the speech recognition system. However, the adaptive optimizations may cause the systems to operate outside of design parameters that impact safety considerations. Also, raw patient data may include sensitive information, and the collection and distribution of such data may raise privacy concerns.

Implementations of the present disclosure can provide means to alleviate some of the aforementioned disadvantages. Some implementations of the present disclosure provide a secure AI enabled wearable sensor platform designed to utilize different sensors to detect onset of infection growth, clots from stenosis, and changes in the local area occurring due to improper maintenance. Some implementations of the present disclosure provide subjective assessments for whether an AVF is fine or whether an AVF is at risk. Some implementations of the present disclosure alert a clinic or an emergency contact to intervene when the AVF is at risk. As such, the intervention can come at a timely fashion to prevent further damage to the AVF. The patient's AVF can, therefore, be monitored more frequently, even continuously, as opposed to every four weeks in conventional methods.

Implementations of the present disclosure can provide systems and devices (for example, a secure AI enabled wearable sensor platform) that use machine learning to continuously track and monitor one or more sensors distributed near a patient's fistula. In some implementations, the described systems and devices use an AI assisted feedback control system (sometimes referred to as an analog front end in this specification) to optimize sensor response characteristics for a specific target objective. In some implementations, the described systems and devices use an AI assisted sensor sampling rate to optimize power usage for specific operational goals, track and monitor sensor performance, and track and monitor calibration data. In some implementations, the described systems and methods use AI assisted signal processing to provide suggestions for a medical professional (for example, a physician overseeing dialysis treatment) that may be used by the medical professional to optimize treatment prescription data. In some implementations, the described systems and devices can perform AI assisted system root-cause analysis and/or determine sensor degradation status, and can include supervisory features to limit AI control features to be within original factory validated/acceptance states.

The types of sensors used can be mapped to the natural biological sensors that may have become degraded or ineffective due to physiological/medical abnormalities. For example, the one or more sensors can include, among others, one or more of the following: a temperature sensor, pulse oximetry (SpO2) sensor, blood pressure sensor, blood flow rate sensor, phosphorous sensors, sodium sensors, and potassium sensors. Implementations of the described systems and devices can periodically determine the health of the access point (fistula) by determining one or more health parameters associated with the fistula, such as optimal transfer rate, likelihood of infection, and so forth. Implementations of the described systems and devices can determine one or more health parameters associated with a natural balanced corporeal operational state of the fistula that may have a vital impact on dialysis transfer efficiency (for example, natural body temperature, blood pressure, SpO2, erythropoietin level, blood flow rate, and so forth). Implementations of the present disclosure can determine the corporal operational state while the patient is undergoing dialysis, or when the patient is not undergoing dialysis.

Implementations of the present disclosure can use secure hash algorithms (SHAs), such as SHA-256 algorithms, to provide secure key based modification (and/or ability to enable/disable) of operational/functional features of AI supervisory functions, calibration algorithms, data/signal processing algorithms, and boundary conditions for the machine learning system to maintain deterministic operational stability of the entire platform. SHA-256 based multimode encoded key algorithms can be used to authenticate each device to a unique patient, patient's treatment data with respect to each dialysis treatment (and dialysis machine), while protecting patient specific data (utilizing traceable security keys and signature metadata) and maintaining safe operating limits. Implementations of the present disclosure can also manage and deploy fault tolerant SHA-256 based “multimode encoded security keys” (e.g., traceable keys) that facilitate remote software updates, remote configuration management, and remote secure data transfer and management.

Implementations of the present disclosure can provide one or more of the following functionalities. Infection growths near patient access points can be monitored and prevented. Clots from stenosis can be monitored and detected. Pre-emptive interventions to fix an AVF can be facilitated. Continuous feedback can be provided to patients for self-management, safety triggered “call back message” (for example, when potassium levels exceed safe limits, blood pressure changes exceed the safe limits, etc.), which can provide feedback for the clinic/physician when the patient is away from the dialysis machine. Continuous kidney sensory related data can be provided to a physician during or after dialysis treatments. Furthermore, in some implementations, when a secure AI enabled wearable sensor platform is within the proximity of a security engine platform (which can be installed in dialysis machine), it can receive, after authentication and handshake protocols, raw sensor data (while away from dialysis and in-dialysis), processed sensor data for suggestions to the physician (for example, indicating key characteristics, trends, etc.), sensor system signature metadata (for example, per request from the Security Engine Platform), or a combination of them, among others.

In the following description, for the purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the present disclosure. It will be apparent, however, that the present disclosure may be practiced without these specific details. In other instances, well-known structures and devices are shown in block diagram form in order to avoid unnecessarily obscuring the present disclosure.

In the drawings, specific arrangements or orderings of schematic elements, such as those representing devices, modules, instruction blocks, and data elements, are shown for ease of description. However, it should be understood by those skilled in the art that the specific ordering or arrangement of the schematic elements in the drawings is not meant to imply that a particular order or sequence of processing, or separation of processes, is required. Further, the inclusion of a schematic element in a drawing is not meant to imply that such element is required in all embodiments or that the features represented by such element may not be included in or combined with other elements in some embodiments.

Further, in the drawings, where connecting elements, such as solid or dashed lines or arrows, are used to illustrate a connection, relationship, or association between or among two or more other schematic elements, the absence of any such connecting elements is not meant to imply that no connection, relationship, or association can exist. In other words, some connections, relationships, or associations between elements are not shown in the drawings so as not to obscure the disclosure. In addition, for ease of illustration, a single connecting element is used to represent multiple connections, relationships, or associations between elements. For example, where a connecting element represents a communication of signals, data, or instructions, it should be understood by those skilled in the art that such element represents one or multiple signal paths (e.g., a bus), as may be needed, to affect the communication.

Reference will now be made in detail to embodiments, examples of which are illustrated in the accompanying drawings. In the following detailed description, numerous specific details are set forth in order to provide a thorough understanding of the various described embodiments. However, it will be apparent to one of ordinary skill in the art that the various described embodiments may be practiced without these specific details. In other instances, well-known methods, procedures, components, circuits, and networks have not been described in detail so as not to unnecessarily obscure aspects of the embodiments.

Several features are described hereafter that can each be used independently of one another or with any combination of other features. However, any individual feature may not provide any of the advantages discussed above or may provide one of the advantages discussed above. Although headings are provided, data related to a particular heading, but not found in the section having that heading, may also be found elsewhere in this description.

FIG. 1 shows a perspective view of a hemodialysis system 100 that includes a hemodialysis machine 102 connected to a disposable blood component set 104 that partially forms a blood circuit. During hemodialysis treatment, an operator connects arterial and venous patient lines 106, 108 of the blood component set 104 to a patient. The blood component set 104 includes an air release device 112, which contains a self-sealing vent assembly that allows air but does not allow liquid to pass. As a result, if blood passing through the blood circuit during treatment contains air, the air release device 112 will vent the air to atmosphere.

The blood component set 104 is secured to a module 130 attached to the front of the hemodialysis machine 102. The module 130 includes the blood pump 132 capable of circulating blood through the blood circuit. The module 130 also includes various other instruments capable of monitoring the blood flowing through the blood circuit. The module 130 includes a door that when closed, as shown in FIG. 1, cooperates with the front face of the module 130 to form a compartment sized and shaped to receive the blood component set 104. In the closed position, the door presses certain blood components of the blood component set 104 against corresponding instruments exposed on the front face of the module 130.

The operator uses a blood pump module 134 to operate the blood pump 132. The blood pump module 134 may include a display window, a start/stop key, an up key, a down key, a level adjust key, and an arterial pressure port. The display window displays the blood flow rate setting during blood pump operation. The start/stop key starts and stops the blood pump 132. The up and down keys increase and decrease the speed of the blood pump 132. The level adjust key raises a level of fluid in an arterial drip chamber. The controls may exist as physical buttons/switches/actuators/etc. or wholly within a graphical user interface on a touch screen accessible display 118 or some combination thereof.

The hemodialysis machine 102 further includes a dialysate circuit formed by the dialyzer 110, various other dialysate components, and dialysate lines connected to the hemodialysis machine 102. Many of these dialysate components and dialysate lines are inside the housing 103 of the hemodialysis machine 102 and are thus not visible in FIG. 1. During treatment, while the blood pump 132 circulates blood through the blood circuit, dialysate pumps (not shown) circulate dialysate through the dialysate circuit.

A dialysate container 124 is connected to the hemodialysis machine 102 via a dialysate supply line 126. A drain line 128 and an ultrafiltration line 129 also extend from the hemodialysis machine 102. The dialysate supply line 126, the drain line 128, and the ultrafiltration line 129 are fluidly connected to the various dialysate components and dialysate lines inside the housing 103 of the hemodialysis machine 102 that form part of the dialysate circuit. During hemodialysis, the dialysate supply line 126 carries fresh dialysate from the dialysate container 124 to the portion of the dialysate circuit located inside the hemodialysis machine 102. As noted above, the fresh dialysate is circulated through various dialysate lines and dialysate components, including the dialyzer 110, that form the dialysate circuit. As will be described later, as the dialysate passes through the dialyzer 110, it collects toxins from the patient's blood. The resulting spent dialysate is carried from the dialysate circuit to a drain via the drain line 128. When ultrafiltration is performed during treatment, a combination of spent dialysate (described below) and excess fluid drawn from the patient is carried to the drain via the ultrafiltration line 129.

The dialyzer 110 serves as a filter for the patient's blood. The dialysate passes through the dialyzer 110 along with the blood, as described above. A semi-permeable structure (e.g., a semi-permeable membrane and/or semi-permeable microtubes) within the dialyzer 110 separates blood and dialysate passing through the dialyzer 110. This arrangement allows the dialysate to collect toxins from the patient's blood. The filtered blood exiting the dialyzer 110 is returned to the patient. The dialysate exiting the dialyzer 110 includes toxins removed from the blood and is commonly referred to as “spent dialysate.” The spent dialysate is routed from the dialyzer 110 to a drain.

A drug pump 192 also extends from the front of the hemodialysis machine 102. The drug pump 192 is a syringe pump that includes a clamping mechanism configured to retain a syringe 178 of the blood component set 104. The drug pump 192 also includes a stepper motor configured to move the plunger of the syringe 178 along the axis of the syringe 178. A shaft of the stepper motor is secured to the plunger in a manner such that when the stepper motor is operated in a first direction, the shaft forces the plunger into the syringe, and when operated in a second direction, the shaft pulls the plunger out of the syringe 178. The drug pump 192 can thus be used to inject a liquid drug (e.g., heparin) from the syringe 178 into the blood circuit via a drug delivery line 174 during use, or to draw liquid from the blood circuit into the syringe 178 via the drug delivery line 174 during use.

The hemodialysis machine 102 includes a security engine platform 105-1 that includes a control unit 101 (e.g., a processor) configured to receive signals from and transmit signals to the display 118, the control panel 120, and a communication module 107 (e.g., a near field communication (NFC) transceiver). The control unit 101 can also communicate with a network, server (e.g., an Internet server), or a master design and test platform (for example, as described later in this specification). In some implementations, the control unit 101 controls operating parameters of the hemodialysis machine 102 based at least in part on signals received by the display 118, the control panel 120, and the communication module 107.

The display 118 presents a user interface that may include vital signs of a patient, operational parameters of the dialysis treatment, and controls associated with the hemodialysis process. For example, the operational parameters may include ultrafiltration parameters, blood pump rate and information associated with the dialysate, hematocrit alert levels, blood pressure alarm limits, medicine infusion parameters, etc. The display 118 may include a touch screen through which the operator may interact and control the hemodialysis machine 102. For example, the operator may input various treatment parameters associated with the hemodialysis process.

The communication module 107 is configured to detect and communicate with an authenticated short-range wireless device—for example, an ID card 105, smart card, mobile device (for example, a smartphone belonging to a patient of medical care professional), and/or one of the wearable secure AI enabled sensor platforms described in this specification—when the device is within its wireless communication range. The communication module 107 is capable of communicating (for example, through multimode encoded security keys) information associated with the patient to the control unit 101. The communication module 107 can send a signal to the control unit 101 indicating that the patient is no longer present.

The control unit 101 is capable of receiving information (for example, through multimode encoded keys) associated with the identity of the patient from the communication module 107 and comparing the information with authentication profiles for multiple users (for example, patients) stored in a database. The database may be stored on a storage device associated with the dialysis system 100 or located at an external storage device (for example a server or a storage device associated with a different dialysis system.) Based on the comparison, the control unit 101 may retrieve the interface configuration profile of the operator or patient. The specific user interface that appears on the display may be based on the retrieved interface configuration profile. For example, the operating parameters and the dialysis controls that constitute the user interface may be determined from the retrieved interface configuration profile. Additionally, identity of the operator or patient, and the identity of a supervisor of the operator may also be displayed.

The control unit 101 is capable of communicating information associated with identity of the patient and a secure AI enabled wearable sensor platform (such as one or more of the wearable sensor platforms described later in this specification). The control unit 101, with a command and control instruction (for example, received from a master design and test platform described in this specification) is capable of adding or modifying an interface and/or operation configuration profile. The control unit 101 may be configured to detect an alert or alarm (used interchangeably herein) in the hemodialysis system 100, and based on the alarm, change the user interface that appears on the display 118 (for example, an alarm specific to a security and/or system anomaly). The alarm may be caused when the value of an operational parameter satisfies a criteria. For example, an alarm may occur when conductivity of the dialysate is above a certain value. This may require immediate attention from a nurse, and therefore the user interface that appears on the display 118 may be changed to a form expected by and/or familiar to the nurse.

FIG. 2 shows a perspective view of a peritoneal dialysis (PD) system 200 that includes a PD cycler (also referred to as a PD machine) 202 seated on a cart 204. The PD cycler 202 includes a housing 206, a door 208, and a cassette interface (not shown) that contacts a disposable PD cassette 212 when the cassette 212 is disposed within a cassette compartment formed between the cassette interface and the closed door 208. A heater tray 216 is positioned on top of the housing 206. The heater tray 216 is sized and shaped to accommodate a bag of dialysate (e.g., a 5 liter bag of dialysate). The PD cycler 202 also includes a display 118 and additional control panel 220 that can be operated by a user (e.g., a patient) to allow, for example, set-up, initiation, and/or termination of a PD treatment.

Dialysate bags 222 are suspended from fingers on the sides of the cart 204, and a heater bag 224 is positioned in the heater tray 216. The dialysate bags 222 and the heater bag 224 are connected to the cassette 212 via dialysate bag lines 226 and a heater bag line 228, respectively. The dialysate bag lines 226 may be used to pass dialysate from dialysate bags 222 to the cassette 212 during use, and the heater bag line 228 may be used to pass dialysate back and forth between the cassette 212 and the heater bag 224 during use. In addition, a patient line 230 and a drain line 232 are connected to the cassette 212. The patient line 230 may be connected to a patient's abdomen via a catheter and may be used to pass dialysate back and forth between the cassette 212 and the patient's peritoneal cavity during use. The drain line 232 may be connected to a drain or drain receptacle and may be used to pass dialysate from the cassette 212 to the drain or drain receptacle during use. The various components that pass dialysate form a dialysate circuit.

The PD system 200, like the hemodialysis system 100, includes a control unit 201, a display 218 and a communication module 207 that communicate with one another, and with a short-range wireless device (for example, the ID card 105 or a smartphone). Similar to the hemodialysis system 100, the PD system 200, using components such as the control unit 201 and the communication module 207, may detect information associated with the identity of an operator or patient, and change the user interface that appears on the display 218 based on an interface configuration profile. The PD system 200 allows an operator or patient to add or modify the interface configuration profile that may be stored in a storage device associated with the PD system 200, or an external storage device (for example, a server). Furthermore, the control unit 201 of the PD system 200 may change the user interface that appears on the display 218 based on an alarm in the PD system 200.

FIG. 3A shows a block diagram of a secure AI enabled wearable sensor platform 304, according to one or more implementations of the present disclosure. The wearable sensor platform 304 includes two or more sensors 312-1 to 312-N (hereinafter collectively referred to as the sensors 312). The sensors may take measurements in connection with a medical area, site or point of interest 306, for example, an access point for an AVF. For example, using the center of the medical point (e.g. access point) 306 for orientation, each of the sensors 312 is located a distance dx from the access point 306 at a direction ?x. The sensors 312 generate sensor data by measuring conditions at specific locations relative to the location of the medical point 306. It is noted that the system described herein is discussed principally in connection with medical sensor platform worn on the arm, e.g. in connection with monitoring a vascular access site; however, the wearable sensor platform of the system described herein may be appropriately utilized in connection with other medical areas and points of interest for a patient, and may be configured to be wearable in locations and manners other than on the arm.

The sensors 312 can include two or more of pressure sensors, temperature sensors, oxygen saturation SpO2 sensors, color sensors (RGB color or XYZ spectral sensors), infrared (IR) sensors, ultrasonic time-of-flight sensors, force sensors like accelerometers, among others. For example, the sensors 312 can include a pulse oximetry (SpO2) sensor that uses a red light source, an infrared light source, photo detectors, and a probe to measure oxygen saturation of a patient's blood. The sensors 312 can include a combination of an RGB Color sensors, XYZ spectral sensors, and infrared sensors that can facilitate determining one or more of: Blood CO2 saturation levels, potassium (K) concentration and quantity, Phosphate (PO3) concentration and quantity, possibility of Anemia, possibility of Fistula Infection (e.g., Fistula health, transfer efficiency, and so forth), and other health parameters (e.g., hormone levels).

The sensors 312 can include MEMS ultrasonic time-of-flight sensors to facilitate determining one or more of: blood flow rate, blood vessel health (e.g., determine degradation amounts for veins, arteries, capillaries, and so forth), and blood vessel transfer efficiency. The sensors 312 can include temperature sensors to facilitate determining one or more of the following: correlations between infection and a healthy access point and patient body responses during and after dialysis. The sensors 312 can include MEMS pressure sensors to facilitate determining one or more of: heart beat rate, blood pressure, correlations between blood flow rate and blood pressure (e.g., in conjunction with MEMS ultrasonic time-of-flight sensors), and heart condition (e.g., in conjunction with MEMS ultrasonic time-of-flight sensors).

In some implementations, two or more of the sensors 312 are tunable/adaptable to measure a desired characteristic, in which the two or more tunable/adaptable sensors 312 are tuned to specific wavelengths appropriate for measuring the desired characteristic of a particular patient. For example, at least some of the sensors 312 and light sources can be tuned to wavelengths appropriate for measuring the potassium levels of a particular patient's blood based on information learned about the particular patient, and then be tuned to wavelengths appropriate for measuring the phosphorous levels of the particular patient's blood based on information learned about the particular patient. The tuning/adapting of sensors to a particular patient is described later in this specification.

The wearable sensor platform 304 includes an analog front end (AFE) 314 for conditioning and processing the sensor data generated. Signal conditioning by the analog front end 314 can include preparing signal levels of sensor data of each of sensors 312 so that analogous data can be compared. Signal conditioning can include applying a Nyquist filter to the sensor data. Signal processing of the sensor data by the analog front end 314 can include using tunable hardware filters capable of extracting stable data of interest with minimal noise. Signal processing can include signal-noise-ratio optimization. For example, RGB IR optical sensors can be packaged with an RGB IR light source. During design of the sensors (for example, during a algorithm and hardware design phase at a master design and test platform as described later in this specification), a testing process can be used that simulates a patient's skin surface. For each target of interest (for example, sodium, potassium, SPO2, etc.), there can be a corresponding spectral frequency response. For a spectral response, signal to noise ratio (SNR) may depend on gain, bandwidth, nature of scattering (reflecting) surface, movements (skin) of the reflecting surface, intensity of light source, ambient light, distance from the reflecting medium, electronic noise (from circuitry), ground noise from power source, surrounding electromagnetic noise induction, and so forth. Based on these factors, outputs of the sensors can be baselined.

Signal processing can include applying one or more tracking algorithms to the conditioned sensor data. The tracking algorithms can be adapted to a particular characteristic of interest. For example, a spectrometer sensor can be tuned to detect potassium based on baseline data and dynamic data collected from the patient through the wearable sensor platform 304. Tuning and tracking algorithms can be defined specific to a sensor and a measured objective (which is independent of patient specific data). Baseline reference data can be created using a master design and test platform by inducing several (or all) possible real-life scenarios/variables. For example, a spectral sensor utilizing absorption/scattering property can be burdened with noise artifacts from skin color, skin movements, and so forth. AI-machine learning systems at the master design and test platform can have tunable algorithms and can use ideal spectral response data as baseline input, and then determine all possible conditions that can affect SNR. AI-machine learning algorithms can be used to create a recognizable higher confidence level to track and measure a desired measurement and/or characteristics of interest (signature metadata). The baseline algorithm and data can create a reference profile for the measurement of interest. When the wearable senor platform 304 is on a patient, the sensor data can be received by an artificial intelligence (AI) machine learning system 340 of the wearable sensor platform 304, and the AI machine learning system 340 can create a signature metadata. The AI machine learning system 340 can be used to compare the sensor data with the baseline data and provide appropriate control parameters to control the analog front end 314 to get the sensor data that can be close to the baseline data (signature/profile) created using the Master Design and Test Platform. This continuous real-time learning and continuous tuning to get optimized data (with optimized SNR) can provide predictive/deterministic measurements, increasing processing time while reducing power consumption (for the machine learning process).

Signal processing can also include one or more of applying sensor response/pattern reorganization algorithms to the conditioned sensor data and applying predictive algorithms to the conditioned sensor data. The analog front end 314 can include tunable filters that facilitate spectral response processing that can be adapted to a particular patient. The analog front end 314 can track the sensors and calibrate the sensors based on data obtained during tracking. The analog front end 314 can algorithmically control the sensors 312 to tune one or more of the sensors to a particular patient such that desired characteristics of the particular patient can be measured, as described previously (e.g., tune an RGB sensor to a particular range of wavelengths so that the sensor data can be used to measure a potassium concentration of the patient). In some implementations, the analog front end 314 applies sensor response/pattern reorganization and predictive algorithms to the sensor data. As will be explained later in this specification, the algorithms can be designed such that they do not infringe upon specified rules and boundary conditions.

The wearable sensor platform 304 includes a processor 316 which performs further preparation of the sensor data for transmission. The processor 316 can be an ultra-low power processor, a field programmable gate array (FPGA), a complex programmable logic device (CPLD), and so on.

The secure AI enabled wearable sensor platform 304 (sometimes referred to as a wearable sensor platform 304 in this specification) includes a wireless communications interface 318 (sometimes referred to as a communications interface 318 or interface 318 in this specification). The communications interface 318 allows communication with one or more external devices, such as a security engine platform residing on a dialysis system (as described previously with respect to FIGS. 1-2. The wearable sensor platform 304, after a successful authentication with a security engine platform, can send status and monitoring data prepared by the security engine platform. The wearable sensor platform 304 can also receive configuration updates, software updates, and calibration updates from the security engine platform. The interface 318 can includes radio frequency transceivers. Examples of interfaces 318 include radio transceivers compatible with several Wi-Fi standards, cellular modems, and so forth.

The wearable sensor platform 304 includes a display 320. The display 320 can include one or more light emitting diode (LED) lights for providing visual indications. The LED lights can indicate that the wearable sensor platform 304 is functioning/not functioning properly and/or the access point 306 is/isn't at risk. In some embodiments, since functioning properly is expected most of the time, the LED lights only turn on to indicate that something is wrong; this way, the LED lights do not consume energy during normal operation. The display 320 can also include one or more auditory indicators like a speaker that beeps when the wearable sensor platform 304 is either not functioning properly or when the access point 306 is at risk. In some embodiments, the display 320 provides a visual and/or auditory indication when the wearable sensor platform 304 first comes online, when the wearable sensor platform 304 is proximate to the dialysis machine (e.g., when the wearable sensor platform 304 detects the presence of the security engine platform by handshake protocols), and when blood flow rate sensor data profiles match to flow rate data during a dialysis treatment.

The wearable sensor platform 304 includes a power source 322 for providing power to the different components described. In some embodiments, the power source 322 is a battery or a rechargeable battery.

The wearable sensor platform 304 includes a security enforcer 323. Generally, the security enforcer 323 is capable of facilitating the updating of the AI machine learning system 340 using a key-based authentication scheme, guard-banding operational/control limits governed by the AI machine learning system 340, providing data (e.g., raw sensor data and process data) to a physician (e.g., via a security engine platform residing in a dialysis system) to assist the physician in deriving quicker and effective decisions regarding a patient and treatments (e.g., prescriptions). The security enforces is also generally capable of autonomously facilitating warning/messaging decisions when any of the data measured by the sensors 312 triggers a critical safety state (e.g., blood pressure reading exceeds safe low and high limits).

The security enforcer 323 is programmable hardware, e.g., an FPGA or a CPLD that serves one or more functions. In some implementations, the security enforcer 323 performs platform management, power management, and security enforcement. For platform management, the security enforcer 323 manages system level critical tasks for the wearable sensor platform 304. Platform management can include ensuring sensor data integrity and operational stability control. For example, the security enforcer 323 receives data from the sensor 312-1. The data from the sensor 312-1 is appended with an authentication code where the security enforcer 323 checks this authentication code against a code from the sensor 312-1 to determine whether the data came from the specific sensor.

Platform management can also include monitoring safety critical status signals and critical signal integrity. Safety critical status signals are signals assigned to represent safe operating limits of the wearable sensor platform 304. An example includes power monitoring signals that define status of the power source 322. The power monitoring signals can indicate a remaining power level where the security enforcer 323 determines whether the remaining power level is adequate to provide trustable information from the sensors 312. For example, if remaining power level falls below a predefined level, then the security enforcer 323 enforces safety protocols. The safety protocols include operational safety protocols and/or human safety protocols. An operational safety protocol can involve storing present data, setting a warning indicator (e.g., blinking orange light from a tri-color LED of the status indicator 320), reducing sampling rate for the sensors 312, and reducing signal processing rate at the analog front end 314 and/or the processor 316. In some implementations, the security enforcer 323 performs human operational control, which can include sounding an alarm via speakers in the display 320 to check for a short circuit that can create high current and may cause burning.

In some embodiments, signal integrity of safety critical status signals cannot be trusted to enforce correct safety protocols. The security enforcer 323 determines whether the signal has too much noise susceptibility. Signals with too much noise susceptibility include, for example, unstable signals due to susceptibility to changes in ambient temperature and unstable signals due to electromagnetic compatibility (EMC) compliances. The security enforcer 323 also monitors abnormal loading of configuration and calibration data. In some implementations, authentication of the security enforcer 323 is required to modify the configuration and calibration data.

Platform management can also include emergency control enforcing for safety critical parameters and functionality. Platform management can also include sensor calibration status check with respect to stored baseline data. Platform management can also include control and retention of factory calibration data, sensor functional/operational configuration data, and/or the wearable sensor platform 304 functional/operational configuration data. Platform management can further include system alarm and safe operating limits management. Platform management can also include safeguarding boot code of the security enforcer 323 from corruption from one or more forms of unauthorized/undesired changes.

Power management of the security enforcer 323 involves power sequencing and reset processing, power monitoring, power management and optimizations for longer time-between-recharge and for performance optimizations, and power emergency controls. Power emergency controls include safety protocol management and safe management of sensor data when available operating power is below a safe operating threshold, as discussed above.

Security enforcement of the security enforcer 323 includes cybersecurity guard/compliance enforcement for secure and safe operation of hardware and software of the wearable sensor platform 304. Security enforcement includes security key management for master authentication and encryption and decryption engines for secure signal and/or data processing, transfer, and storage. In master authentication, a security key is transferred from a manufacturing site of the wearable sensor platform 304 (after factory acceptance tests) to authenticate and enable secure engine functionality. At the manufacturing site, device operational configuration files can be loaded, calibrated, and tested before approval for customer delivery. Security enforcement also includes threat monitoring and critical event logging, e.g., logging of safety and security events, logging of remote accesses to the wearable sensor platform, and so on. Security enforcement can also include a safety and security mode (including internal and external data management and communication/data transfer).

In some implementations, security enforcement includes receiving a multi-mode encoded key (e.g., traceable key) from an outside source (e.g., a standalone board residing in a dialysis machine). In some implementations, the multi-mode encoded key is designed by the manufacturing site (for example, a master design and test platform as described later in this specification). The multi-mode encoded key can correspond with a secure hash algorithm (SHA), such as an SHA256 algorithm. The security enforcer 323 can process the multimode encoded keys to generate an identification key particular to the patient using the wearable sensor platform 304, and use the identification key to encrypt the raw sensor data captured by the sensors 312, such that the encrypted raw sensor data is traceable to the particular patient. As will be described later, the encrypted raw sensor data can be transmitted to an outside source (e.g., a security engine platform as described in this specification) for further processing. The security enforcer 323 can also receive the multimode encoded keys from a security engine platform to authenticate the wearable sensor platform 304 to receive software updates from the security engine platform (e.g., updates to machine learning algorithms, rules and boundary conditions, and so forth, as explained later in more detail).

Security enforcement by the security enforcer 323 also includes supervising the AI machine learning system 340. For example, the security enforcer 323 can guard-band operational/control limits governed by the AI machine learning system 340, as described later in more detail. The guard-banding can, for example, ensure that algorithmic modifications to the sensors 312 and/or the analog front end 314 do not cause these components to operate outside of their pre-validated state (e.g., in accordance with validation performed by a manufacturer/Master Design and Test Platform, as described later in more detail).

The AI machine learning system 340 applies rules/goals based (e.g., baseline rules/goals that are designed and validated by a mater design and test platform) machine learning techniques to train the machine learning model 340a that, when applied to input data, generates indications of whether the input data items have an associated property or properties. As part of the training of the machine learning model 340a, the AI machine learning system 340 can form a training set of input data by identifying a positive training set of input data items that have been determined to have the property in question, and, in some implementations, forms a negative training set of input data items that lack the property in question.

The AI machine learning system 340 extracts values from the input data of the training set, the values being variables deemed potentially relevant to whether or not the input data items have the associated property or properties (e.g., defined by the rules/goals). An ordered list of the features for the input data is herein referred to as the feature vector for the input data. In some implementations, the AI machine learning system 340 applies dimensionality reduction (e.g., via linear discriminant analysis (LDA), principle component analysis (PCA), learned deep features from a neural network, or the like) to reduce the amount of data in the feature vectors for the input data to a smaller, more representative set of data.

In some implementations, the AI machine learning system 340 uses supervised machine learning (for example, at a master design and test platform to create baseline profile data for each sensors and corresponding measurements) to train the machine learning model 340a with the feature vectors of the positive training set and the negative training set serving as the inputs. Different machine learning techniques—such as linear support vector machine (linear SVM), boosting for other algorithms (e.g., AdaBoost), neural networks, logistic regression, naïve Bayes, memory-based learning, random forests, bagged trees, decision trees, boosted trees, or boosted stumps—are used in some implementations. The machine learning model 340a, when applied to the feature vector extracted from the input data item, outputs an indication of whether the input data item has the property in question, such as a Boolean yes/no estimate, a scalar value representing a probability, a vector of scalar values representing multiple properties, or a nonparametric distribution of scalar values representing different and not a priori fixed numbers of multiple properties, which may be represented either explicitly or implicitly in a Hilbert or similar infinite dimensional space.

In some implementations, a validation set is formed of additional input data, other than those in the training sets, which have already been determined to have or to lack the property in question. The AI machine learning system 340 applies the trained machine learning model 340a to the data of the validation set to quantify the accuracy of the machine learning model 340a. Common metrics applied in accuracy measurement include: Precision=TP/(TP+FP) and Recall=TP/(TP+FN), where precision is how many the machine learning model 340a correctly predicted (TP or true positives) out of the total it predicted (TP+FP or false positives), and recall is how many the machine learning model 340a correctly predicted (TP) out of the total number of input data items that did have the property in question (TP+FN or false negatives). The F score (F-score=2*PR/(P+R)) unifies precision and recall into a single measure. In some implementations, the machine learning system 340 iteratively re-trains the machine learning model 340a until the occurrence of a stopping condition, such as the accuracy measurement indication that the model 340a is sufficiently accurate, or a number of training rounds having taken place.

In some implementations, the machine learning model 340a includes a neural network. In some implementations, the neural network includes a convolutional neural network. The machine learning model 340a can include other types of neural networks, such as recurrent neural networks, radial basis function neural networks, physical neural networks (e.g., optical neural network), and so forth.

The machine learning model 340a is configured to modify, based on learned features particular to the patient using the wearable sensor platform 304, the algorithms used by the analog front end 314 to perform signal processing. For example, based on learned characteristics of the patient, the machine learning model 340a can tune the tunable filters of the analog front end 314 to a particular patient to facilitate adaptable spectral response processing. The machine learning model 340a can also update the user adaptable tracking algorithms and SNR optimization algorithms of the analog front end 314, based on learned characteristics of the patient. The machine learning model 340a can implement rules and boundary conditions set forth by the security enforcer 323 to ensure that any algorithmic updates do not cause the wearable sensor platform 304 to infringe validated states as set forth by a manufacturer/Master Design and Test Platform (described later). The machine learning model 340a can include a safe-operation look-up table that consists of operating parameters and limits related to basic safety and essential performances. The safe-operation look-up table can be used to create rules and constraints for the AI-machine learning system to protect against any system changes that will induce non-validated states or conditions for the platform 304. The machine learning model 340a can use function enable keys to facilitate changes or modifications, or enable/disable functionality with respect to the AI-machine learning system.

FIG. 3B shows a block diagram of an environment 300 for monitoring a vascular access, according to one or more implementations of the present disclosure. The environment includes and Master Design and Test (MDT) platform 370 (which can be integrated with a security engine platform 361), a dialysis system 360, and the wearable sensor platform 304 described previously with reference to FIG. 3A.

In general, the MDT platform 370 provides remote software updates and remote diagnostics for the wearable sensor platform 304 and the security engine platform 361 of the dialysis system 360. The MDT platform 370 is also generally capable of providing deep root-cause analysis for the security engine 361 of the dialysis system 360 and wearable sensor platform 304 (e.g., any root-cause analysis that is beyond the capabilities of the security engine platform 361 and wearable sensor platform 304). The MDT platform 370 is generally capable of providing the backbone for deep trend analysis and comparative (snapshot) studies among several security engines 361, along with continuous design improvements, rule/goal tables, and so forth.

The MDT platform 370 is configured to generate and validate the multi-mode encoded keys described previously with reference to FIG. 3A. The MDT platform 370 is configured to provide model based design and development environments for the design of hardware, software, and algorithms used with the wearable sensor platform 304 and the security engine platform 361 of the dialysis system 360, as well as the optimization, verification, and validation of such algorithms.

The dialysis system 360 can be similar to the dialysis systems 100, 200 described previously with reference to FIGS. 1-2. The dialysis system 360 includes the security engine platform 361, which can be a stand-alone board that resides in dialysis equipment of the dialysis system 360. The security engine platform 361 can authenticate the wearable sensor platform 304 and its functions. The security engine 361 can receive the multi-mode encoded keys from the MDT platform 370, and provide the keys to the wearable sensor platform 304 (e.g., when the wearable sensor platform 304 is within a close range due to the patient undergoing dialysis treatment). The security engine platform 361 can use the multi-mode encoded keys to authenticate and encrypt/decrypt data transferred between the security enforcer 323 of the wearable sensor platform 304 and the security engine 361. The security engine platform 361 is configured to perform software updates and lifecycle management of the wearable sensor platform 304. The security engine platform 361 is configured to receive one or more of: the captured raw sensor data from the wearable sensor platform 304 and the processed sensor data from the wearable sensor platform 304. As indicated previously, the captured raw sensor data and/or the processed sensor data can be encrypted by the wearable sensor platform 304 (e.g., using the security enforcer 323) with a multi-mode encoded keys. In some implementations, the security engine platform 361 can include corresponding multi-mode encoded keys which can be used to decrypt the received data. The security engine 361 can transmit the received data to the MDT platform 370 before or after encryption. In some implementations, the MDT platform 370 includes corresponding multi-mode encoded keys for decrypting the data. As indicated previously, the processed sensor data can include intelligent information. The security engine 361 can provide the raw captured data and the processed sensor data, along with encrypted prescription data representing prescriptions used by the dialysis system 360 to perform treatment and performance data indicating how the dialysis system 360 performed for a specific prescription, to a physician (e.g., to an electronic device of the physician or through a user interface of the dialysis machine).

FIG. 4 shows a flowchart of a method 400 for monitoring the health of a patient, according to one or more implementations of the present disclosure. For illustrative purposes, the method 400 is described as being performed by the wearable sensor platform 304, as described previously with reference to FIG. 3A. The method 400 includes one or more of: receiving physiological measurements (block 402), determining differential readings (block 404); determining if the differential readings exceed an alarm threshold (block 405); determining if physiological measurements approximately equal a nominal value (block 406); determining if the physiological measurements deviate from the nominal value by a threshold value (block 407); determining trend information (block 408); determining whether to generate an alert based on the trend information (block 410); and generating an alert (block 409), which may include displaying the alert on an appropriate display, as further discussed herein.

At block 402, data representing physiological measurements captured at different sensor locations of the wearable sensor platform 304 are received by the analog front end 314.

At block 404, the analog front end 314 provides normalized raw data to the AI machine learning system 340 to determine differential readings between the different sensor locations based on the physiological measurements. The analog front end incorporates mutual distances between the sensor locations and distances between the access point 306 and the sensor locations with the physiological measurements to determine the differential readings between the different sensor locations.

At block 405, in some implementations, the analog front end 314, in conjunction with the AI machine learning system 340, determines whether the differential readings exceed an alarm threshold. The alarm threshold can be determined by the machine learning model 340a based on information learned about the user of the wearable sensor platform 304. For example, the machine learning model 340a can learn a standard baseline reference (sometimes referred to as a static threshold in this specification) temperature near a fistula (for example, from a plurality of patients) and identify a new threshold for a particular patient based on sensor data collected over time by the wearable sensor platform 304 (sometimes referred to as a dynamic threshold in this specification). The machine learning model 340a can then use both the static threshold and the dynamic threshold to identify undesired states near the fistula. If, for example, an alarm threshold is exceeded, then at block 409, the security enforcer 323 generates an alarm. Generating an alarm can include causing the display 320 to exhibit visual cues (e.g., blinking lights) and/or acoustical cues (e.g., beeping sounds). Generating an alarm can include alerting an external systems/devices, such as the dialysis system 360, a mobile device of a physician, or a mobile device of the user of the wearable sensor platform 304.

At block 406, the analog front end 314 determines whether the physiological measurements approximate a nominal value. In some implementations, the physiological measurements approximate the nominal value when the physiological measurements are within a deviation percentage of 1% of the nominal value. In some implementations, the nominal value and the deviation percentage is determined by the machine learning model 340a based on information learned about the patient. A state where the physiological measurements approximate the nominal value may be referred to as an equilibrium state.

At block 407, the analog front end 314 determines whether the physiological measurements deviate from the nominal value by a threshold. The threshold can be determined by the machine learning model 340a based on information learned about the patient. In some implementations, if the analog front end 314 and the AI machine learning system 340 determine that the physiological measurements deviate from the nominal value by a threshold, then at block 309, the security enforcer 323 generates an alarm, as discussed previously.

In some implementations, at block 408, if the physiological measurements deviate from the nominal value by the threshold, the analog front end 314 and the AI machine learning system 340 determine trend information of the physiological measurements. Trend information includes tracking the physiological measurements for a specified time period to determine whether the physiological measurements are steady, are rising, or are falling. A state where trend information is determined can be called a caution state. The specified time period can be determined by the machine learning model 340a based on information learned about the user. In some implementations, the on/off sampling rate of a sensor is based on rates of change (and trend/variance results from the AI machine learning system 340) in data within a defined time window, power management rules and constraints (e.g., due to the wearable nature of the platform 304).

At 410, the analog front end 314 determines whether to generate an alert based on the trend information. For example, if the trend information shows certain measurements rising faster than a threshold rate of change, then the security enforcer 323 can generate an alert. The threshold rates of change can be user specific and determined by the machine learning model 340a based on information learned about the user.

As illustrated in FIG. 4, in some implementations, the analog front end 314 can receive physiological measurements at block 402 and then proceed to block 406 to determine whether the physiological measurements approximately equate the nominal value.

FIG. 5 shows a secure AI enabled wearable sensor platform 500, implemented as a sleeve worn on an arm of a patient, having temperature sensors for differential temperature measurements, according to one or more implementations of the present disclosure. In some implementations, the wearable sensor platform 304 discussed previously with reference to FIG. 3A includes some or all aspects of the wearable sensor platform 500. The wearable sensor platform 500 includes an analog front end 514, a security enforcer 523, a processor 516, a machine learning system 540, and a display 520, which can be substantially similar to corresponding components of the wearable sensor platform 304 discussed previously with reference to FIG. 3A.

The wearable sensor platform 500 includes a first temperature sensor 504-1, a second temperature sensor 504-2, and a third temperature sensor 504-3, each placed at different locations from an access point 502 and communicatively coupled to the analog front end 514. In some implementations, the first temperature sensor 504-1 and the second temperature sensor 504-2 are placed at opposite directions from the access point 502. In the illustrated example, the access point 502 is located on one side of the arm, and the third temperature sensor 504-3 is located on the opposite side of the arm. The third temperature sensor 504-3 location is chosen to be as far away from the access point 502 but equidistant to the first and second sensors 504-1, 504-2.

The spatial distance from the access point 502 to the first temperature sensor 504-1 is d1. The spatial distance from the access point 502 to the second temperature sensor 504-2 is d2. The spatial distance from the first temperature sensor 504-1 to the third temperature sensor 504-3 is d3, and the spatial distance from the second temperature sensor 504-2 to the third temperature sensor 504-3 is also d3.

In some implementations, having three readings from three different locations provides an area equivalent calculation for determining qualities of the access point 502. The different readings at the different locations when combined allow for performing trend analysis, developing standard deviations, and developing temperature data profiles and tracking algorithms. The multiple sensor configuration can provide deterministic and highly-reliable system decisions based on triple-redundant sensor data from two or more sensors. This allows decision based voting where a majority of sensors agreeing can be used as a strong bias towards determining whether a minority of sensors are faulty. In some implementations, the temperature sensors 504-1, 504-2, 504-3 sample body temperature, in degrees Celsius, at their respective locations once every minute.

The wearable sensor platform 500 can be configured to perform the method 400 discussed previously with reference to FIG. 4. For example, at block 402, the temperature sensors 504-1, 504-2, 504-3 can collect sensor data at a predetermined sampling rate (e.g., every minute) and sends the sensor data after a prescribed amount of time to the analog front end 514. The sampling rate can be determined by the AI machine learning system 540 based on information learned about the user, and governed by the security enforcer 523 such that the sampling rate is more particular to the user but stays within a validated/safe state.

At block 404, using the spatial distances d1, d2, and d3, and the temperature measurements in the sensor data of block 402, the analog front end 514 can then determine differential temperature readings between the different sensors as provided in Eqs. 1-3. d1 and d2 can be 3-5 cm from the access point 502.



Eq. 1 is the differential temperature reading between the first temperature sensor 504-1 and the second temperature sensor 504-2, where T1 and T2 are temperature measurements from the first and second temperature sensors 504-1, 504-2, respectively. Eq. 2 is the differential temperature reading between the first and third temperature sensors 504-1, 504-3, where T3 is the temperature measurements from the third temperature sensor 504-3. Eq. 3 is the differential temperature reading between the second and third temperature sensors 504-2, 504-3. T1, T2 and T3 are temperature measurements over a time period (t.stop-t.start). Thus, Eqs. 1-3 find the difference between average temperature readings of the first and second temperature sensors 504-1, 504-2, the first and third temperature sensors 504-1, 504-3, and the second and third temperature sensors 504-2, 504-3, respectively.

From the temperature measurements T1, T2 and T3 and the differential temperature readings of Eqs. 1-3, at block 406, the analog front end 514 determines whether each temperature measurement approximates a nominal temperature Tnom. In some implementations, temperature measurements T1, T2 and T3 should be within a deviation percentage of 1% of the nominal temperature Tnom. However, the deviation percentage can be determined by the machine learning system 540 as described previously. In the illustrated implementations, when T1, T2 and T3 are within 1% of Tnom, the wearable sensor platform 500 maintains an equilibrium state. When any of T1, T2 or T3 is not within 1% of Tnom, the wearable sensor platform 500 transitions to a caution state. The caution state signifies that the body temperature is either too high or too low. As described previously, Tnom can be determined by the machine learning system 540 based on information learned about the user.

In some implementations, at block 405, the differential temperature readings of Eqs. 1-3 can be used to determine whether the differential temperature readings exceed an alarm threshold. As discussed previously, the alarm threshold Talarm, can be determined by the machine learning system 540 based on information learned about the user. In some implementations, when Eq. 1, 2 or 3 is nonzero and the magnitude of Eq. 1, 2 or 3 evaluated exceeds an alarm threshold Talarm, then an alarm is generated at block 409. Eq. 4 summarizes this condition.


In Eqs. 1-4, the following definitions apply:

Talarm is Tnom×1%
Tnom is normal body temperature computed within 24 hours and can be determined by the machine learning system 540 as described previously.
Measurements T1, T2 and T3 made with the temperature sensors provide values for body temperature Tbody that can be used to determine a body temperature trend bandwidth Tbody±dT.

At block 408, the client device determines trend information using the temperature measurements if the physiological measurements are determined at block 407 to deviate from Tnom by more than 1%. In the event that the wearable sensor platform 500 transitions to a caution state, the temperature sensors 504-1, 504-2, 504-3 will change their sampling rates of temperature measurements T1, T2 and T3 from a slower sampling rate to a faster sampling rate. For example, the sampling rate can change from one temperature measurement per minute to one temperature measurement per second. This faster sampling rate is maintained for a predefined period (e.g., maintained for 5 to 30 minutes). The faster sampling rate and predefined period can be determined by the machine learning system 540 based on information learned about the user (for example, baseline biological reference data, such as blood flow rates). The determined sampling rates and predefined period can be governed by the security enforcer 523 such that the rates/periods stay within a validated/safe state. The temperature measurements obtained during this predefined period are used to determine a temperature trend Ttrend.

In some implementations, when using average temperatures, a faster sampling rate also shortens the averaging time. For example, if sampling at one temperature measurement per second, then average temperatures can be obtained for every 30 seconds, thus the temperature trend Ttrend is determined with average temperature values obtained every 30 seconds for a 5 to 30 minute period. The temperature trend Ttrend is a relationship describing change in the different temperature values obtained every 30 seconds.

At block 410, the analog front end 514 analyzes Ttrend to determine whether to generate an alert. For example, when Ttrend indicates steady temperature levels and body temperature is greater than Tnom+Tnom×1%, then the analog front end 514 can send a message to the security enforcer 523. The security enforcer 523 can then generate an alert (block 409) by, for example, sending a message to a clinic treating the user and/or activating the display 520. When Ttrend indicates rising temperatures during the predefined period (e.g., 5 minutes or 30 minutes) and body temperature is greater than Tnom+Tnom×1%, then the security enforcer 523 can generate an alert. When Ttrend indicates falling temperatures and body temperature is less than Tnom-Tnom×1%, then the security enforcer 523 can generate an alert.

The following example certain aspects of the method 400 as previously described with respect to the wearable sensor platform 500 of FIG. 5. The distances d1, d2, and d3 are constants from the view of the placements of temperature sensors 504-1, 504-2, 504-3. Algorithms applied to temperature readings from the temperature sensors 504-1, 504-2, 504-3 therefore account for variables of T (temperature) and t (time).

For example, t=tstart is an arbitrary starting value (in time) where temperature T is sampled, and t=tstop is a time where sampling of temperature T is finished. Each temperature sensor 504-1, 504-2, 504-3 continuously reads from tstart to tstop, and the resulting temperature reading is the integration of T from tstart to tstop

In some implementations, for saving power, the difference between tstart and tstop is small or configurable/adjustable. For example, the duration tstart-tstop can be 1 millisecond. A period of no measurements can then ensue. Afterward, the temperature sensors 504-1, 504-2, 504-3 repeat sampling temperature T for another time duration (e.g., tstart2-tstop2=1 ms).

In some implementations, two temperature readings are taken between tstart and tstop. For example, when the duration is a 1 ms duration, a temperature measurement T1 is made at tstart and a temperature measurement T1.1 is made at tstop. From T1 and T1.1, different cases can be used to determine whether the patient has an infection or elevated body temperature. Two cases are described with respect to the T1 and T1.1 measurements:

Case 1: T1-T1.1=0

Case 1 implies that T1 is equal to T1.1, thus there is no change in temperature during the 1 ms duration. Either T1 or T1.1 is compared against normal human body temperature (baselined as the normal body temperature of the patient). If either T1 or T1.1 is over 1% higher than the normal body temperature, then the patient can be deemed to have a fever.

Case 2: T1>T1.1 or T1<T1.1

Case 2 implies that there is a temperature gradient or a local heat source which signals a possibility of an infection or some kind of an anomaly. Depending on the sensor being used, the distance dx can be used to interpret location of the heat source. For example, when the first temperature sensor 504-1 is placed closer to the access point 502, then elevated temperatures measured at the first temperature sensor 504-1 indicate that the infection is close to the first temperature sensor 504-1. When there is an infection, the human body reacts and elevates body temperature near the location of the infection. Since in FIG. 5 there are three temperature sensors 504-1, 504-2, 504-3 at different distances from the access point 502, an algorithm can be used to combine temperature readings from the different temperature sensors 504-1, 504-2, 504-3 to obtain a more deterministic output (independent of the ambient temperature, and the reading uncertainty from a single temperature sensor).

The previous example showed use of two measurement points in a 1 ms duration. In some embodiments, tstart-tstop is long enough to obtain more than two temperature measurements, so each temperature measured can be stored in memory and used for further analysis. The integral functions will begin to look like a simplified form of the Stokes equation and the Fundamental theorem of Calculus.

In some implementations of the method 400 of FIG. 4, using the exemplary configuration of sensors illustrated in FIG. 5, temperature readings from the sensors 3504-1, 504-2, 504-3 are sampled to provide data points, T1, T2, and T3, respectively. Sampling frequency (e.g., sensor on/off duty cycle and sensor sequencing rate) may be dependent on power consumption, battery life, and machine learning. The following discussion uses the first temperature sensor 504-1 as an example, but similar measurements and analyses may be performed in tandem at the second and third temperature sensors 504-2, 504-3.

For example, at block 402, the first temperature sensor 504-1 samples a first reading of T1. The first reading can be a single read, i.e., a single sample, or can be a multiple read, e.g., an average of three fast samples for eliminating body/ambient noise. A multiple read and a single read both provide a single value for the first reading. Design and validation of several algorithms (including baseline references, safety limits, safe operating conditions/constraints, baseline reference rate of change for any sensor reading that can be an indicator for unsafe health conditions) can be completed at a master design and test platform. Each set of sensor algorithms can have two distinct operational functional groups. One groups can use the baseline reference data and AI machined learned data (corresponding rule/goals). The other group can be used to learn and create (for example, using the AI machine learning system) psychological profile for each patient.

At block 406, the first reading is then compared to a nominal body temperature.

At block 407, if the first reading is greater than the nominal body temperature, then at block 408, T1 is sampled for a longer time to determine a temperature trend. Using the temperature trend, then fever (abnormal temperature) can be determined using a rising, falling, or steady state trend of the temperature trend.

That is, after comparison of the first reading with the nominal body temperature, then the sampling time is increased to obtain more temperature readings so that the temperature trend can be determined. The temperature trend will give a direction of change as well as a rate of change so as to know whether the temperature is increasing, decreasing, or remaining steady during the longer sampling time. The longer sampling time or duration is identified above as the duration tstart-tstop. The duration can be short to obtain a single T value or long to obtain multiple T values.

At block 410, based on the trend information, an alert can be generated.

FIG. 6 shows a wearable sensor platform 600 having temperature sensors for gradient temperature measurements, according to one or more implementations of the present disclosure. In some implementations, the wearable sensor platform 304 discussed previously with reference to FIG. 3A includes some or all aspects of the wearable sensor platform 600. The wearable sensor platform 600 includes an analog front end 614, a security enforcer 623, a processor 616, a machine learning system 640, and a display 620, which can be substantially similar to corresponding components of the wearable sensor platform 304 discussed previously with reference to FIG. 3A.

The wearable sensor platform 600 includes a first temperature sensor 604-1, a second temperature sensor 604-2, and a third temperature sensor 604-3, each placed at different locations from an access point 602. In an embodiment, the first and second temperature sensors 604-1, 604-2 are placed at different distances but in a same direction from the access point 602. The access point 602 is located on one side of the arm, and the third temperature sensor 604-3 is located on the opposite side of the arm. For example, anterior compartment placement vs. posterior compartment placement on the forearm. As illustrated in FIG. 6, the placement of the first and second temperature sensors 604-1 and 604-2, can indicate whether there is an infection and help determine a location of the infection.

The spatial distance from the access point 602 to the first temperature sensor 604-1 is d1. The spatial distance from the access point 602 to the second temperature sensor 604-2 is d2. The spatial distance from the first temperature sensor 604-1 to the third temperature sensor 604-3 is d3, and the spatial distance from the second temperature sensor 604-2 to the third temperature sensor 604-3 is d4.

In some implementations, the temperature sensors 604-1, 604-2, 604-3 sample body temperature, in degrees Celsius, at their respective locations once every minute. The wearable sensor platform 600 can perform the method 400 described previously with reference to FIG. 4. For example, the temperature sensors 604-1, 604-2, 604-3 can collect sensor data every minute and sends the sensor data to the analog front end 614 at block 402.

At block 404, differential temperature readings can be determined using the temperature measurements and the spatial distances in a similar manner as described previously with reference to FIG. 5. Differential temperature readings between the first and second temperature sensors 604-1, 604-2 follow Eq. 1, and differential temperature readings between the first and third temperature sensors 604-1, 604-3 follow Eq. 2. Similarly, differential temperature readings between the second and third temperature sensors 604-2, 604-3 are determined by Eq. 5 since distance between the first and third temperature sensors 604-1 and 604-3 and the distance between the second and third temperature sensors 604-2, 604-3 are not equal.

?
t
.
start



At block 406, to remain in the equilibrium state, as previously discussed with respect to FIG. 5, temperature measurements T1, T2 and T3 should be, in some implementations, within 1% of the nominal temperature Tnom. When T1, T2 and T3 are within 1% of Tnom, the wearable sensor platform 600 maintains the equilibrium state. When any of T1, T2 or T3 is not within 1% of Tnom, the wearable sensor platform 600 can transition to a caution state. The caution state may signify that the body temperature is too high or too low. As discussed previously with respect to FIG. 5, the temperature measurements T1, T2 and T3 can be average temperature values and not instantaneous measurements.

Similar to the implementation illustrated in FIG. 5, the differential temperature readings can also be used to generate an alarm when Eq. 4 is true. That is, at block 405, if the differential readings exceed an alarm threshold, then at block 409, an alarm is generated. Based on the outcome of block 407, blocks 408 and 410 can be performed in a similar manner where a temperature trend Ttrend is determined and an alarm is generated using Ttrend and the body's temperature relative to the normal temperature Tnom.

Although discussed separately, implementations provided in FIG. 5 and FIG. 6 can be combined to determine both that an infection is present and also a location of the infection. That is, in various embodiments, four or five sensors can be arranged in a wearable sensor platform to obtain both differential and gradient temperature measurements.

FIG. 7 shows a wearable sensor platform 700 having pressure sensors, according to one or more implementations of the present disclosure. In some implementations, the wearable sensor platform 304 discussed previously with reference to FIG. 3A includes some or all aspects of the wearable sensor platform 700. The wearable sensor platform 700 includes an analog front end 714, a security enforcer 723, a processor 716, a machine learning system 740, and a display 720, which can be substantially similar to corresponding components of the wearable sensor platform 304 discussed previously with reference to FIG. 3A.

The wearable sensor platform 700 includes a first pressure sensor 704-1 and a second pressure sensor 704-2, each placed at different locations from an access point 702. The access point 702 is located on one side of the arm, and the second pressure sensor 704-2 is located on the opposite side of the arm. Pressure measurements from the wearable sensor platform 700 can be used to determine whether the access point 702 is at risk of stenosis and/or hematoma. Furthermore, since pressure changes in blood vessels more rapidly than temperature, detection of access point bulging can be performed quickly.

In some implementations, the pressure sensors 704-1, 704-2 are placed above the patient's skin and measures pulsing below the skin. The pulsing can be used to estimate blood vessel pressure once every minute in, e.g., mmHg. The spatial distance from the access point 702 to the first pressure sensor 704-1 is d1. The spatial distance from the first pressure sensor 704-1 to the second pressure sensor 704-2 is d2. In some implementations, the spatial distance d2 is greater than d1.

The wearable sensor platform 700 can perform the method 400 discussed previously with reference to FIG. 4. For example, the pressure sensors 704-1, 704-2 can collect sensor data every minute and send the sensor data to the analog front end 614 client device at block 402.

At block 404, the differential pressure reading between the first pressure sensor 704-1 and the second pressure sensor 704-2 is determined via Eq. 6, where P1 and P2 are pressure measurements obtained from the first and second pressure sensors 704-1, 704-2, respectively. In some implementations, average values of P1 and P2 are used in Eq. 6 instead of instantaneous values.

?
t
.

In Eq. 7, the following definitions apply:

Palarm is either Palarm1 or Palarm2 depending on which sensor's measurements are being considered;
Palarm1=Pnom1×1%;
Palarm2=Pnom2×1%;
Pnom1 in is normal body pressure near fistula, computed within 24 hours pressure trend analysis;
Pnom2 in is normal body pressure near fistula, computed within 24 hours pressure trend analysis
Measurements P1 and P2 made with pressure sensors provide values for pressure at specific sites and can be used to determine a body pressure trend bandwidth which is defined as P1 or 2±dP1 or 2, where ±dP1 or 2 is unique to each person, computed over a larger period (e.g., >24 hours).

At block 408, the analog front end 714 determines trend information using the pressure measurements if the physiological measurements are determined at block 407 to deviate from Pnom1 or Pnom2 by more than 1%. In the event that the wearable sensor platform 700 transitions to a caution state, the pressure sensors 704-1, 704-2 will change their sampling rates for pressure measurements P1 and P2 from a slower sampling rate to a faster sampling rate. For example, the sampling rate can change from one pressure measurement per minute to one pressure measurement per second. The faster sampling rate is maintained for a predefined period of time (e.g., maintained for 5 minutes). As discussed previously, the sampling rates and period of time can be determined by the machine learning system 740 based on information learned about the user, and controlled by the security enforcer 723 such that they do not operate outside of a validated/safe state. The pressure measurements obtained during this predefined period of time are used to determine pressure trends Ptrend1 and Ptrend2.

At block 410, the analog front end 714 analyzes each of Ptrend1 and Ptrend2 to determine whether to generate an alert. Each of Ptrend1 and Ptrend2 will be referred to as Ptrend and each of Pnom1 and Pnom2 will be referred to as Pnom for the following description. Superscripts can be added to obtain trend analysis for each pressure sensor's measurement. When Ptrend indicates steady pressure levels and body pressure is greater than Pnom+Pnom×1%, then the security enforcer 723 generates an alert. When Ptrend indicates rising pressure levels during the predefined period of time and body pressure is greater than Pnom+Pnom×1%, then security enforcer 723 generates an alert. When Ptrend indicates falling pressure levels and body pressure is less than Pnom-Pnom×1%, then the security enforcer 723 generates an alert.

Although the implementation illustrated in FIG. 7 includes two pressure sensors, more than two pressure sensors can be included in the wearable sensor platform 700. Placement of sensors can be a combination of that shown for the temperature sensors illustrated in FIGS. 5-6. Multiple sensors being used can provide advantages that were previously described in the temperature sensor setting (e.g., decision based voting, high-reliable system decisions based on triple-redundant sensor data, and so forth).

Furthermore, although a dialysis state with respect to temperature was not described with reference to FIGS. 5-6, in some implementations, temperature measurements during dialysis are also tracked. That is, temperature readings from the temperature sensors can be compared against a Tdialysis to determine whether to exit an equilibrium state. Furthermore, a diffTnom and a diffTdialysis can be obtained for each of Eqs. 1-3 during a normal state and during a dialysis state, respectively. Furthermore, a diffTn-d can be defined as diffTnom-diffTdialysis. Since the use of measurement metrics in the normal state and the dialysis state is already described above with respect to pressure, similar descriptions are omitted with respect to temperature.

FIG. 8 shows a wearable sensor platform 800 having color sensors, according to one or more implementations of the present disclosure. In some implementations, the wearable sensor platform 304 discussed previously with reference to FIG. 3A includes some or all aspects of the wearable sensor platform 800. The wearable sensor platform 800 includes an analog front end 814, a security enforcer 823, a processor 816, a machine learning system 840, and a display 80, which can be substantially similar to corresponding components of the wearable sensor platform 304 discussed previously with reference to FIG. 3A.

The wearable sensor platform 800 includes a first and second color sensor 804-1, 804-2, each being placed at different locations from an access point 802. The access point 802 is located on one side of the arm, and the second color sensor 804-2 is located on the opposite side of the arm. As discussed previously, the color sensors 804-1, 804-2 can be tuned to different frequencies for measuring physiological characteristics of the patient. For example, the color sensors 804-1, 804-2 can include a combination of RGB color sensors, XYZ spectral sensors, and infrared sensors that can facilitate measuring one or more of the following: SpO2, blood CO2 saturation levels, potassium (K) concentration and quantity, Phosphate (PO3) concentration and quantity, possibility of anemia, possibility of fistula infection (e.g., Fistula health, transfer efficiency, and so forth), and other health parameters (e.g., hormone levels). The tuning of each of the color sensors 804-1, 804-2 can be determined by the AI machine learning system 840 and be based on information learned about the patient, as described previously.

For illustrative purposes, the following description of the wearable sensor platform 800 will assume that the color sensors 804-1, 804-2 are tuned for facilitating SpO2 measurements. SpO2 measurements from the wearable sensor platform 800 can be used to check for onset of anemia around the area of the access point 802. SpO2 measurements can also provide information on level of iron deficiency there is in the body.

The spatial distance from the access point 802 to the first color sensor 804-1 is d1. The spatial distance from the first color sensor 804-1 to the second color sensor 804-2 is d2. In some implementations, the spatial distance d2 is greater than d1. The wearable sensor platform 800 can perform the method 400 described previously with reference to FIG. 4. For example, the color sensors 804-1, 804-2 can collect sensor data every minute and send the sensor data after a prescribed amount of time to the analog front end 814.

At block 804, the differential SpO2 reading between the first color sensor 804-1 and the second color sensor 804-2 is determined via Eq. 8, where SPO21 and SPO22 are SpO2 measurements obtained from the first and second color sensors 804-1, 804-2, respectively.



The differential SpO2 reading determined via Eq. 8 can produce different values under a normal state, when a patient is not undergoing dialysis, and under a dialysis state, when the patient is undergoing dialysis. As such, Eq. 8 can be used to obtain differential SpO2 readings under both the normal state and the dialysis state, and the difference in differential SpO2 readings between the normal state and the dialysis state can be used to determine whether physiological measurements are approximately within a normal range at block 406.

In a normal state, SPO21=SPO22=SPO2nom, so at block 406, the analog front end 814 determines whether the SpO2 measurements SPO21 and SPO22 approximate a nominal blood oxygen saturation SPO2nom. In a dialysis state, SPO21=SPO22=SPO2dialysis, so at block 406, the analog front end 814 determines whether the SpO2 measurements SPO21 and SPO22 approximate a dialysis blood oxygen saturation SPO2dialysis. Applying Eq. 8 to SpO2 measurements obtained in a normal state will provide diffSPO2nom, and applying Eq. 8 to SpO2 measurements obtained in the dialysis state will provide diffSPO2dialysis. The difference between diffSPO2nom and diffSPO2dialysis is diffSPO2n-d. In some implementations, diffSPO2n-d is bounded by a factor dSPO2n-d, where dSPO2n-d is unique to each person and the state of the fistula or access point 802 computed over a larger period (e.g., >24 hours) trend analysis.

In some implementations, there can be different nominal SpO2 values for the color sensors 804-1, 804-2. That is, SPO21 values are checked to see whether they approximate SPO2nom1 and SPO22 values are checked to see whether they approximate SPO2nom2. Each of SPO2nom1 and SPO2nom2 can have a deviation of dSPO2nom which sets a bounds for “approximately” as used previously in comparison. dSPO2nom is unique to each person and the state of the fistula or access point 802 computed over a larger period (e.g., >24 hours) trend analysis. When SPO21 is within 1% of SPO2nom1 and when SPO22 is within 1% of SPO2nom2, then the wearable sensor platform 800 maintains an equilibrium state. When any of these is untrue, then the wearable sensor platform 800 transitions to a caution state. The caution state signifies that one or more body SpO2 measurements is too high or too low. 1% is used as an example here, but other percentage deviations, e.g., 2%, from nominal pressure can be used as limits.

At block 405, the differential SpO2 readings of Eq. 8 can be used to determine whether to generate an alarm. In an embodiment, when Eq. 8 is nonzero and the magnitude of Eq. 8 evaluated exceeds an alarm threshold, then an alarm is generated. Note that since each SpO2 sensor has its own nominal value, SPO2nom1 or SPO2nom2, there are two alarm thresholds to consider SPO2alarm1 and SPO2alarm2. Eq. 9 summarizes conditions for generating an alarm.

When
?
(
?
t
.
start
t
.
stop
d

?

alarm

In Eq. 9, the following definitions apply:

SPO2alarm is either SPO2alarm1 or SPO2alarm2 depending on which sensor's measurements are being considered
SPO2alarm1=SPO2nom1×1%
SPO2alarm2=SPO2nom2×1%
SPO2nom1 is normal body SpO2 level near fistula, computed within 24 hours SpO2 trend analysis
SPO2nom2 is normal body SpO2 level near fistula, computed within 24 hours SpO2 trend analysis
Measurements SPO21 and SPO22 made with color sensors provide values for blood oxygen saturation levels at specific sites and can be used to determine a body SpO2 trend bandwidth which is defined as SPO21 or 2±dSPO21 or 2, where ±dSPO21 or 2 is unique to each person, computed over a larger period (e.g., >24 hours).

At block 408, the analog front end 814 determines trend information using the SpO2 measurements if the physiological measurements are determined at block 407 to deviate from SPO2nom1 and SPO2nom2 by more than 1%. In the event that the wearable sensor platform 800 transitions to a caution state, the first and second sensors 804-1, 804-2 will change their sampling rates for SpO2 measurements SPO21 and SPO22 from a slower sampling rate to a faster sampling rate. For example, the sampling rate can change from one SpO2 measurement per minute to one SpO2 measurement per second. The faster sampling rate is maintained for a predefined period of time (e.g., maintained for 5 to 30 minutes). The SpO2 measurements obtained during this predefined period of time are used to determine blood oxygen saturation trends SPO2trend1 and SPO2trend2.

At block 410, the analog front end 814 analyzes each of SPO2trend1 and SPO2trend2 to determine whether to generate an alarm. Each of SPO2trend1 and SPO2trend2 will be referred to as SPO2trend and each of SPO2nom1 and SPO2nom2 will be referred to as SPO2nom for the following description. Superscripts can be added to obtain trend analysis for each sensor's measurement. When SPO2trend indicates steady SpO2 levels and body SpO2 is greater than SPO2nom+SPO2nom×1%, then an alarm is generated. When SPO2trend indicates rising SpO2 levels during the predefined period of time and body SpO2 is greater than SPO2nom+SPO2nom×1%, then an alarm is generated. When SPO2trend indicates falling SpO2 levels and body SpO2 is less than SPO2nom-SPO2nom×1%, then an alarm is generated.

Although described separately, each of the implementations illustrated in FIGS. 5-8 is combinable with other illustrated implementations. For example, different types of sensors can be provided in a wearable sensor platform to provide information on infection and hematoma using temperature and pressure. Furthermore, other types of sensors like accelerometers and force sensors can be used to determine whether the patient is putting too much weight or is moving an arm too quickly. Information obtained by these sensors can be used to alert emergency contacts so that they can inform the patient on how to better be aware of posture so as to not generate such alerts. Embodiments of the disclosure provide a way of educating patients on proper maintenance of fistulas to combat any lack of awareness of potentially dangerous behavior or improper maintenance on the patient's part.

Implementations of the disclosure can be used to monitor peritoneal dialysis access points or to monitor other surgical grafts. A wearable sensor platform according to some implementations of the disclosure can be provided as a band especially when all sensors are placed on a same side.

In an example, a patient puts on a wearable sensor platform according to implementations of the disclosure. The wearable sensor platform is fitted with sensor straps that prevent the sensors from moving once in place. The system then turns on automatically after the sensor straps are fixed. The sensors provide pressure and temperature data which are used to determine whether the access point being monitored is healthy. A physician/clinic is alerted by the wearable sensor platform if there is an anomaly. In some implementations, the wearable sensor platform is worn during dialysis to inform a physician on how to, for example, adjust blood flow rate of the dialysis machine based on performance of the access point.

FIG. 9 shows a block diagram of an environment 900 including a factory module 902, a supervisor module 904, and a plurality of operations modules 906-908, according to one or more implementations of the present disclosure. The factory module 902 may be provided at a point of manufacture of a wearable sensor platform (such as the platform 304 described previously with reference to FIG. 3A) or a dialysis system (such as the dialysis systems 100, 200, 360 described in this specification). In some implementations, the MDT platform 370 described previously with reference to FIG. 3B includes the factory module 902. The supervisor module 904 is provided on a dialysis system (such as the dialysis systems 100, 200, 360 described in this specification) and the operation modules 906-908 are provided on a wearable sensor platform (such as the platforms 304, 500, 600, 700, 800 described in this specification). In some implementations, the security engine platform 361 described previously with reference to FIG. 3B includes the supervisor module 904, and at least one of the operations modules 906-908 can encompass the security enforcer 323, machine learning system 340, the analog front end 314, and the sensors 312 described previously with reference to FIG. 3A. Each of the operations modules 906-908 can encompass different wearable sensor platforms. As indicated previously, the factory module 902 provides, in a secure and non-repudiable manner, limits/constraints to the supervisor module 904 that the supervisor module 904 uses to provide limits/constraints in a secure and non-repudiable manner to provide for adaptive optimizations/state changes to each of the operation modules 906-908.

The supervisor module 904 constrains adaptive optimizations of the operation modules 906-908 provided by machine learning systems to maintain each of the operation modules 906-908 in a state that is consistent with operating in a safe and effective manner and consistent with prior regulatory approval for each of the modules 906-908. For example, the supervisor module 904 can provide limits/constraints to a security enforcer (such as the security enforcer 323 discussed previously with reference to FIG. 3A) such that it can determine if the state of a particular operation module is acceptable after having been altered using machine learning (e.g., artificial intelligence).

In some implementations, a set of constraints for at least one of the operation modules 906-908 may be determined when a wearable sensor platform is manufactured (or sometime after manufacture in connection with maintenance). In some implementations, different specific units of the same model of wearable sensor platform can use different constraints due to calibration differences and/or different deployment scenarios. The factory module 902 provides constraint data to the supervisor module 904 and provides a cryptographic one-way hash of the constraint data. In some implementations, the cryptographic one-way hash is an SHA256 hash. In some implementations, other appropriate cryptographic one-way hash algorithms are used. The factory module 902 can generate a private/public cryptographic key pair and may provide the public key to the supervisor module 904. In some implementation, the public key may be provided in a digital certificate that is signed by a trusted authority under the X.509 PKI standard. The factory module 902 may then provide to the supervisor module 904 a digital signature of the cryptographic one-way hash of the constraint data. The supervisor module 904 may then verify the source of the digitally-signed cryptographic one-way hash of the constraint data using the public key of the factory module 902 and may then authenticate the received constraint data using the one-way hash value. This mechanism can prevent repudiation of the constraint data since the only possible source is the factory module 902, assuming that the private key has not been disclosed to any party outside the factory module 902. For each of the operation modules 906-908, the factory module 902 can generate a multi-mode encoded key pair, in which the private key is sent to and obtained by the supervisor module 904, and the public key is sent to the supervisor module 904 and transmitted to one of the operation modules 906-908. The multi-mode encoded key pair mechanism can be used by the supervisor module 904 to authenticate data received from each of the operation modules 906-908.

FIG. 10 shows a block diagram of a method 1000 performed in connection with transferring state constraint data from a factory module to a supervisor module, according to one or more implementations of the present disclosure. In some implementations, the factory module 902 performs the method 1000 in connection with providing the constraint data to the supervisor module 904. The method 1000 includes generating constraint data (block 1002), determining one-way hash value (block 1004), digitally signing one-way hash value (block 1006), and providing data and values to supervisor module (block 1008).

At block 1002, the factory module 902 determines the constraint data. Generally, constraints placed on a particular operation module can be determined based on initial values used for the particular module, as well as a determination of any limits needed for providing safe and effective operations. Some of the constraints may be determined empirically based on clinical/testing experience. Some of the constraints can be based, at least in part, on regulatory considerations, where values of some operational parameters are not allowed to go beyond values used for a particular operation module in connection with previously receiving regulatory approval.

At block 1004, a one-way hash value (e.g., SHA256) of the constraint data is calculated. At block 1006, MDT provides digital signatures from foundation security keys to the factory system, to authenticate and validate a wearable sensor platform. The platform 902 provides a digital signature of the one-way hash value. At block 1008, the factory module 902 provides the supervisor module 904 with the constraint data, the one-way hash value of the constraint data, and the digital signature of the one-way hash value of the constraint data. The method 1000 may be performed when a wearable sensor platform is first manufactured or in connection with subsequent maintenance after the wearable sensor platform has been operational.

As indicated previously, the supervisor module 904 can have a private key and a corresponding public key that may be used for the secure and non-repudiable transfer of data (e.g., constraint data) to the operation modules 906-908. Just as with transfer of data from the factory module 902 to the supervisor module 904, data from the supervisor module 904 may be transferred to each of the operation modules 906-908 by first determining a one-way has of the data (using, for example, the SHA256 one-way cryptographic hash mechanism), digitally signing the one-way hash value, and then transferring the data, the one-way hash value, and the digital signature of the one-way hash value. Similarly, each of the operation modules 906-908 may have a private key and a corresponding public key that may be used to provide a secure and non-repudiable data transfer to the supervisor module 904. For example, each of the operation modules 906-908 may use the key pairing for encrypting raw and processed sensor data, and sending the sensor data to the supervisor module 904. The supervisor module 904 may then authenticate each of the operation modules and the data received from each of the operation module 906-908.

FIG. 11A shows a flowchart of a method 1100a performed by an AI machine learning system in conjunction with a security enforcer, according to one or more implementations of the present disclosure. For example, as previously indicated, the machine learning systems described in this specification (such as the machine learning system 340 described previously with reference to FIG. 3A) can perform this method in conjunction with the security enforcers described in this specification (such as the security enforcer 323 described previously with reference to FIG. 3A) to provide constrained adaptive optimization of a corresponding wearable sensor platform. Components that may be adapted by the method 1100a can include sensor settings (such as the frequency and sampling rates of the sensors described in this specification) and processing algorithms used by analog front ends described in this specification (such as nominal values used in previously described algorithms. In general, any operational aspect of a wearable sensor platform may be adapted using machine learning (artificial intelligence). The method 1100a includes assuming an initial state (block 1102), receiving data (block 1104), adjusting a state (block 1106), determining if the state is acceptable (block 1108), and reverting to a previous state (block 1110).

At block 1102, an initial state of a wearable sensor platform is determined. The initial state may be determined at the time of manufacture or may be otherwise input to the system by a supervisor module (discussed in more detail elsewhere herein) or some other external source. For example, the initial state may include an initial value for sensor settings (e.g., an initial sample rate, an initial frequency, and so forth) and/or an initial nominal value (e.g., using a predetermined standard body temperature for as a nominal temperature). In some cases the initial state may be based on the specific operator and/or patient, which may be detected automatically, as described previously.

At block 1104, inputs are provided to the machine learning system. The machine learning system receives appropriate input data for the operations being optimized.

At block 1106 the machine learning system adapts operational parameters/algorithmic parameters according to the input data received. Generally, the specific adaptive optimizations that occur depend upon the input data, the machine learning mechanism that is used (e.g., neural networks), and the specific objective of the machine learning optimizations (e.g., optimize sensor settings).

At block 1108, it is determined if the state of the module is acceptable after the adaptive optimizations have been applied at block 1106. Although adaptive optimizations using artificial intelligence may be useful, it is important that the adaptive optimizations do not cause the system to operate in a way that is potentially unsafe for patients or otherwise detrimental. This is enforced by a security enforcer (such as the security enforcer 323 discussed previously with reference to FIG. 3A), which is discussed previously in more detail.

If it is determined at block 1108 that the new state of the module is not acceptable, then control transfers from block 1108 to block 1110, where the state of the module is restored/reverted to a previous state. The previous state may be either an initial state (i.e., the state set at block 1102, described previously), the state of the module just prior to execution of block 1106, or some other state that is otherwise acceptable. The previous state may be an original validated state. Following block 1110, control transfers back to block 1104 for another iteration. In some implementations, block 1104 also follows directly from block 1108 if it determined that the state of the operation module is acceptable.

FIG. 11B shows a flowchart of a method 1100b performed by a security enforcer in conjunction with a machine learning system, according to one or more implementations of the present disclosure. In some implementations, the method 1100b is performed by the security enforcers described in this specification (such as the security enforcer 323 described previously with reference to FIG. 3A) in conjunction with performing blocks 1108-1110 of the method 1100a discussed previously with reference to FIG. 11A.

The method 1100b includes receiving state data (block 1112), determining if the state data is within constraints (block 1114), sending confirmation (block 1116) and sending reverted state (block 1118).

At block 1112, the security enforcer receives state data for the new state of the operation component being tested (such as the sensors, analog front end, and so forth) from the machine learning system.

At block 1114, the security enforcer examines the state data to determine if any of the operational parameters of the new state are outside any constraints that were provided as part of the constraint data previously received from a security engine (such as the security engine 361 discussed previously with reference to FIG. 3B), which, in some implementations, was received by the security engine from an MDT platform (such as the MDT platform 370 discussed previously with reference to FIG. 3B. If none of the new operational parameters are outside any of the constraints, then control transfers from block 1104 to block 1106 where a confirmation is sent from the security enforcer to the machine learning system. Receipt of the confirmation causes the machine learning system to cause the component to begin operating in the new state. If it is determined at block 1104 that at least one of the new operational parameters are outside the constraints, then control transfers from block 1104 to block 1108 where a data for a previous state is sent to the machine learning system, which then causes the component to revert to the previous state.

FIG. 12 shows a block diagram of an environment 1200 including a wearable sensor platform 1202, a security engine 1206, and a user 1204, according to one or more implementations of the present disclosure. Each of the wearable sensor platforms described in this specification (such as the wearable sensor platform 304 described previously with reference to FIG. 3A) can include the wearable sensor platform 1202. The wearable sensor platform 1202 is coupled to the user 1204 that provides input to the wearable sensor platform 1202 using sensors of the wearable sensor platform 1202. In some situations, the wearable sensor platform 1202 may also be coupled to the security engine 1206 (such as when the patient is undergoing dialysis treatment) that provides baseline data and constraints to a security enforcer 1214 of the wearable sensor platform 1202, as previously described in more detail.

As previously described, the wearable sensor platform 1202 includes an AI machine learning system 1208 that can use one or more artificial intelligence based optimization techniques. The techniques can provide data to a dynamic operational data set 1212 and receive data from the security enforcer 1214. The security enforcers includes mechanisms for constraining critical parameters, as previously described, and may include a control value table, a guard/compliance enforcer, AI specific learning rules, and mechanisms for handling loading and monitoring of dynamic system baseline data. The security enforcer can exchange data with the security engine 1206 when the security engine 1206 is present (for example, using short range wireless communication and secure communications, as described previously). The security enforcer 1214 also receives data from a static operational data set 1216. The static operational data set 1216 includes static (fixed) counterparts for data provided by the dynamic operational data set 1212. In some implementations, both the dynamic operational data set 1212 and the static operational data set 1216 include static and dynamic corresponding data for system specific operating parameters, system calibration data, system configuration data, system alarm limits and safe operating limits, and embedded processor/FPGA boot code. The security engine 1206 provides authentication and facilitates copying data from the static operational data set 1216 to the security enforcer 1214.

The machine learning system 1208 exchange data with a first adaptive control algorithm 1222 and a corresponding first adaptive correction algorithm 1224 and exchange data with a second adaptive control algorithm 1226 and a corresponding second adaptive correction algorithm 1228.

The first adaptive control algorithm 1222 and the first adaptive correction algorithm 1224 may perform this optimization using any one of a number of possible mechanisms, including known optimization mechanisms. However, for safety/regulatory considerations, the machine learning system 1208 is constrained by the security enforcer 1214 to prevent operational parameters from deviating too far from values of the static operational data set 1216. In a similar way, the second adaptive control algorithm 1226 and the corresponding second adaptive correction algorithm 1228 may be used to optimize other operational parameters, such as a sensor's sampling rate.

FIG. 13 is a block diagram of an example computer system 1300 used to provide computational functionalities associated with described algorithms, methods, functions, processes, flows, and procedures described in the present disclosure, according to some implementations of the present disclosure. The illustrated computer 1302 is intended to encompass any computing device such as a server, a desktop computer, a laptop/notebook computer, a wireless data port, a smart phone, a personal data assistant (PDA), a tablet computing device, or one or more processors within these devices, including physical instances, virtual instances, or both. The computer 1302 can include input devices such as keypads, keyboards, and touch screens that can accept user information. Also, the computer 1302 can include output devices that can convey information associated with the operation of the computer 1302. The information can include digital data, visual data, audio information, or a combination of information. The information can be presented in a graphical user interface (UI) (or GUI).

The computer 1302 can serve in a role as a client, a network component, a server, a database, a persistency, or components of a computer system for performing the subject matter described in the present disclosure. The illustrated computer 1302 is communicably coupled with a network 1330. In some implementations, one or more components of the computer 1302 can be configured to operate within different environments, including cloud-computing-based environments, local environments, global environments, and combinations of environments.

At a high level, the computer 1302 is an electronic computing device operable to receive, transmit, process, store, and manage data and information associated with the described subject matter. According to some implementations, the computer 1302 can also include, or be communicably coupled with, an application server, an email server, a web server, a caching server, a streaming data server, or a combination of servers.

The computer 1302 can receive requests over network 1330 from a client application (for example, executing on another computer 1302). The computer 1302 can respond to the received requests by processing the received requests using software applications. Requests can also be sent to the computer 1302 from internal users (for example, from a command console), external (or third) parties, automated applications, entities, individuals, systems, and computers.

Each of the components of the computer 1302 can communicate using a system bus 1303. In some implementations, any or all of the components of the computer 1302, including hardware or software components, can interface with each other or the interface 1304 (or a combination of both), over the system bus 1303. Interfaces can use an application programming interface (API) 1312, a service layer 1313, or a combination of the API 1312 and service layer 1313. The API 1312 can include specifications for routines, data structures, and object classes. The API 1312 can be either computer-language independent or dependent. The API 1312 can refer to a complete interface, a single function, or a set of APIs.

The service layer 1313 can provide software services to the computer 1302 and other components (whether illustrated or not) that are communicably coupled to the computer 1302. The functionality of the computer 1302 can be accessible for all service consumers using this service layer. Software services, such as those provided by the service layer 1313, can provide reusable, defined functionalities through a defined interface. For example, the interface can be software written in JAVA, C++, or a language providing data in extensible markup language (XML) format, or a design language HDL for FPGA and/or model based design language (such as MathWorks). While illustrated as an integrated component of the computer 1302, in alternative implementations, the API 1312 or the service layer 1313 can be stand-alone components in relation to other components of the computer 1302 and other components communicably coupled to the computer 1302. Moreover, any or all parts of the API 1312 or the service layer 1313 can be implemented as child or sub-modules of another software module, enterprise application, or hardware module without departing from the scope of the present disclosure.

The computer 1302 includes an interface 1304. Although illustrated as a single interface 1304 in FIG. 13, two or more interfaces 1304 can be used according to particular needs, desires, or particular implementations of the computer 1302 and the described functionality. The interface 1304 can be used by the computer 1302 for communicating with other systems that are connected to the network 1330 (whether illustrated or not) in a distributed environment. Generally, the interface 1304 can include, or be implemented using, logic encoded in software or hardware (or a combination of software and hardware) operable to communicate with the network 1330. More specifically, the interface 1304 can include software supporting one or more communication protocols associated with communications. As such, the network 1330 or the interface's hardware can be operable to communicate physical signals within and outside of the illustrated computer 1302.

The computer 1302 includes a processor 1305. Although illustrated as a single processor 1305 in FIG. 13, two or more processors 1305 can be used according to particular needs, desires, or particular implementations of the computer 1302 and the described functionality. Generally, the processor 1305 can execute instructions and can manipulate data to perform the operations of the computer 1302, including operations using algorithms, methods, functions, processes, flows, and procedures as described in the present disclosure.

The computer 1302 also includes a database 1306 that can hold data for the computer 1302 and other components connected to the network 1330 (whether illustrated or not). For example, database 1306 can be an in-memory, conventional, or a database storing data consistent with the present disclosure. In some implementations, database 1306 can be a combination of two or more different database types (for example, hybrid in-memory and conventional databases) according to particular needs, desires, or particular implementations of the computer 1302 and the described functionality. Although illustrated as a single database 1306 in FIG. 13, two or more databases (of the same, different, or combination of types) can be used according to particular needs, desires, or particular implementations of the computer 1302 and the described functionality. While database 1306 is illustrated as an internal component of the computer 1302, in alternative implementations, database 1306 can be external to the computer 1302.

The computer 1302 also includes a memory 1307 that can hold data for the computer 1302 or a combination of components connected to the network 1330 (whether illustrated or not). Memory 1307 can store any data consistent with the present disclosure. In some implementations, memory 1307 can be a combination of two or more different types of memory (for example, a combination of semiconductor and magnetic storage) according to particular needs, desires, or particular implementations of the computer 1302 and the described functionality. Although illustrated as a single memory 1307 in FIG. 13, two or more memories 1307 (of the same, different, or combination of types) can be used according to particular needs, desires, or particular implementations of the computer 1302 and the described functionality. While memory 1307 is illustrated as an internal component of the computer 1302, in alternative implementations, memory 1307 can be external to the computer 1302.

The application 1308 can be an algorithmic software engine providing functionality according to particular needs, desires, or particular implementations of the computer 1302 and the described functionality. For example, application 1308 can serve as one or more components, modules, or applications. Further, although illustrated as a single application 1308, the application 1308 can be implemented as multiple applications 1308 on the computer 1302. In addition, although illustrated as internal to the computer 1302, in alternative implementations, the application 1308 can be external to the computer 1302.

The computer 1302 can also include a power supply 1314. The power supply 1314 can include a rechargeable or non-rechargeable battery that can be configured to be either user- or non-user-replaceable. In some implementations, the power supply 1314 can include power-conversion and management circuits, including recharging, standby, and power management functionalities. In some implementations, the power-supply 1314 can include a power plug to allow the computer 1302 to be plugged into a wall socket or a power source to, for example, power the computer 1302 or recharge a rechargeable battery.

There can be any number of computers 1302 associated with, or external to, a computer system containing computer 1302, with each computer 1302 communicating over network 1330. Further, the terms “client,” “user,” and other appropriate terminology can be used interchangeably, as appropriate, without departing from the scope of the present disclosure. Moreover, the present disclosure contemplates that many users can use one computer 1302 and one user can use multiple computers 1302.

FIG. 14 is a schematic diagram 1400 illustrating an example of information flow and process interactions between different components and representative functional modules of a wearable sensor platform as described herein. As illustrated by way of example, the information and interaction exchanges are shown schematically with respect to components of the wearable sensor platform 304 discussed in connection with FIG. 3A, including the sensors 312, security enforcer 323, the sensor AFE 314, and the AI machine learning system 340, and in connection with representations of operations of functional modules, such as a sensor assembly module 1410, a pre-defined controls module 1420, an operational state module 1430, an alert module 1440 and an AI module 1450. It is noted that components of the wearable sensor platform 304 other than that shown in this figure may be used in connection with the described operations and functions of the functional modules, and the components of the wearable sensor platform 304 may perform multiple operations and functions both as part of and in addition to the operations of the described functional modules.

The sensor assembly module 1410 is illustrated schematically in connection with components such as the sensors 312, the security enforcer 323 and the sensor AFE 314, and information and data outputs from those components. The sensor assembly module 1410 further includes representations of functional operations for supervisor module interfacing 1411 (e.g. with the supervisor module 904), AI SHA256 security key management 1412, and predefined & acceptable safe state 1413. The predefined & acceptable safe state 1413 may include processing corresponding to information and configurations for initial operational state, power up state, reset state, configuration state and data acquisition synchronization & initial state of the wearable sensor platform 304.

As constrained by the supervisor module and management using the security key management techniques discussed herein, the predefined controls module 1420 provides operational information and functions that include conditions and processing for rules & limits 1421, critical data/signal lists 1422, acceptable safe state limits 1423 and AI rule tables 1424. Decisions and processes for operational states of the wearable sensor platform 304 are shown in connection with functions of the operational state module 1430, including assuming an initial operational state (block 1431) after activation of the system and sensor system initialization, and receiving sensor data (block 1432). As a result of functioning of the AI module 1450, input to the operational state module 1430 provides for adjusting the operational state (block 1433), determining whether the current operational state is acceptable (block 1434), and determining whether it is necessary to revert to a previous state (block 1435). Alerts or alarms are generated in connection with functioning of the alert module 1440, and which may include generating and/or displaying out-of-state warnings and records 1441.

Functioning of the AI module 1450, using inputs such as received AI control inputs, authorization, and sensor data measured and transmitted by the sensors 312, is enabled by the use of the AI and machine learning algorithms of the AI machine learning system 340 discussed herein. The AI module 1450 functions enable the generating of information and outputs used to adjust on an on-going or continuous basis controls and operational states of the wearable sensor platform 304. The schematically illustrated information flow and interactions between the components and representative functional modules of the wearable sensor platform 304 provide for the secure AI enabled operational adjustment of the wearable sensor platform 304 in connection with that described in detail herein.

FIG. 15 is a schematic illustration showing an example of a connected medical system, for example, a connected health (CH) system 1500, that may be used in connection with network aspects of the wearable sensor platform 304 described herein. The CH system 1500 may include, among other things, a processing system 1505, a CH cloud 1510 and a gateway (CH Gateway) 1520. The processing system 1505 may be a server and/or cloud-based system that processes, compatibility checks and/or formats medical information, including prescription information generated at a clinical information system (CIS) 1504 of a clinic or hospital, in connection with data transmission operations of the CH system 1500. The CH system 1500 may include appropriate encryption and data security mechanisms, like that discussed herein. The CH cloud 1510 may be a cloud-based application or system that serves as a communication pipeline (e.g., facilitates the transfer of data) among components of the CH system 1500 via connections to a network, the Internet.

The gateway 1520 may serve as a communication device facilitating communication among components of the CH system 1500 and other components. In various embodiments, the gateway 1520 is in communication with a dialysis machine 1502, such as a home hemodialysis machine, via a wireless connection 1501, such as a Bluetooth, Wi-Fi and/or other appropriate type of local or short-range wireless connection. The gateway 1520 may also be in connection with the CH cloud service 1510 via a secure network (e.g. Internet) connection. The gateway 1520 is configured to transmit/receive data to/from the CH cloud 1510 and transmit/receive data to/from the dialysis machine 1502. The gateway 1520 may also facilitate communication between peripheral devices, such as a smartphone and/or the wearable sensor platform 304 that may be worn by a patient 10, for example, as worn like one of more of the sleeves implementations discussed in connection with wearable sensor platforms 500, 600, 700, 800 discussed herein.

In an embodiment, during dialysis, e.g. hemodialysis, the dialysis machine 1502 interfaces with the patient 10 at a dialysis access point 11 (e.g., AVF). The dialysis machine 1502 extracts blood from the patient via a blood extraction needle, filters the blood, and reintroduces the blood to the patient via a blood injection needle. That is, during dialysis, the dialysis machine 1502 pumps blood out of the patient via the blood extraction needle, and using tubing, provides the blood to a dialyzer where the blood is filtered. The dialysis machine 1502 then circulates the blood via tubing to the blood injection needle for reintroduction to the patient.

As the patient 10 is undergoing dialysis, the wearable sensor platform 304, using the AI control features and techniques discussed herein, measures conditions of the patient, for example, conditions proximate to the access point 11 and generates sensor data. The wearable sensor platform 304 provides the sensor data and other information to the gateway 1520. The dialysis machine 1502, connected to the gateway 1520 via the wireless connection 1501, also provides treatment information to the gateway 1520. Treatment information can include blood pump rate, dialysate pump rate, type of dialysate used, treatment start time, treatment time elapsed, measured hematocrit, and so on.

The gateway 1520 can receive the data from the wearable sensor platform 304 and from the dialysis machine 1502. In some implementations, the gateway 1520 can provide the sensor data to the dialysis machine 1502 and/or can provide the treatment information and the sensor data to the CH cloud 1510 for further processing and analysis at a remote location, such as the CIS 1504. The gateway 1520 can also receive patient electronic records and prescription information from the CH cloud 1510 and that may be further transmitted, processed and used in connection with operation of the dialysis machine 1502 and/or the wearable sensor platform 304.

Implementations of the subject matter and the functional operations described in this specification can be implemented in digital electronic circuitry, in tangibly embodied computer software or firmware, in computer hardware, including the structures disclosed in this specification and their structural equivalents, or in combinations of one or more of them. Software implementations of the described subject matter can be implemented as one or more computer programs. Each computer program can include one or more modules of computer program instructions encoded on a tangible, non-transitory, computer-readable computer-storage medium for execution by, or to control the operation of, data processing apparatus. Alternatively, or additionally, the program instructions can be encoded in/on an artificially generated propagated signal. For example, the signal can be a machine-generated electrical, optical, or electromagnetic signal that is generated to encode information for transmission to suitable receiver apparatus for execution by a data processing apparatus. The computer-storage medium can be a machine-readable storage device, a machine-readable storage substrate, a random or serial access memory device, or a combination of computer-storage mediums.

The terms “data processing apparatus,” “computer,” and “electronic computer device” (or equivalent as understood by one of ordinary skill in the art) refer to data processing hardware. For example, a data processing apparatus can encompass all kinds of apparatus, devices, and machines for processing data, including by way of example, a programmable processor, a computer, or multiple processors or computers. The apparatus can also include special purpose logic circuitry including, for example, a central processing unit (CPU), a field programmable gate array (FPGA), or an application specific integrated circuit (ASIC). In some implementations, the data processing apparatus or special purpose logic circuitry (or a combination of the data processing apparatus or special purpose logic circuitry) can be hardware- or software-based (or a combination of both hardware- and software-based). The apparatus can optionally include code that creates an execution environment for computer programs, for example, code that constitutes processor firmware, a protocol stack, a database management system, an operating system, or a combination of execution environments. The present disclosure contemplates the use of data processing apparatuses with or without conventional operating systems, for example, LINUX, UNIX, WINDOWS, MAC OS, ANDROID, or IOS.

A computer program, which can also be referred to or described as a program, software, a software application, a module, a software module, a script, or code, can be written in any form of programming language. Programming languages can include, for example, compiled languages, interpreted languages, declarative languages, or procedural languages. Programs can be deployed in any form, including as stand-alone programs, modules, components, subroutines, or units for use in a computing environment. A computer program can, but need not, correspond to a file in a file system. A program can be stored in a portion of a file that holds other programs or data, for example, one or more scripts stored in a markup language document, in a single file dedicated to the program in question, or in multiple coordinated files storing one or more modules, sub programs, or portions of code. A computer program can be deployed for execution on one computer or on multiple computers that are located, for example, at one site or distributed across multiple sites that are interconnected by a communication network. While portions of the programs illustrated in the various figures may be shown as individual modules that implement the various features and functionality through various objects, methods, or processes, the programs can instead include a number of sub-modules, third-party services, components, and libraries. Conversely, the features and functionality of various components can be combined into single components as appropriate. Thresholds used to make computational determinations can be statically, dynamically, or both statically and dynamically determined.

The methods, processes, or logic flows described in this specification can be performed by one or more programmable computers executing one or more computer programs to perform functions by operating on input data and generating output. The methods, processes, or logic flows can also be performed by, and apparatus can also be implemented as, special purpose logic circuitry, for example, a CPU, an FPGA, or an ASIC.

Computers suitable for the execution of a computer program can be based on one or more of general and special purpose microprocessors and other kinds of CPUs. The elements of a computer are a CPU for performing or executing instructions and one or more memory devices for storing instructions and data. Generally, a CPU can receive instructions and data from (and write data to) a memory. A computer can also include, or be operatively coupled to, one or more mass storage devices for storing data.

Computer readable media (transitory or non-transitory, as appropriate) suitable for storing computer program instructions and data can include all forms of permanent/non-permanent and volatile/non-volatile memory, media, and memory devices. Computer readable media can include, for example, semiconductor memory devices such as random access memory (RAM), read only memory (ROM), phase change memory (PRAM), magnetoresistive random access memory (MRAM), static random access memory (SRAM), dynamic random access memory (DRAM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), and flash memory devices. Computer readable media can also include, for example, magnetic devices such as tape, cartridges, cassettes, and internal/removable disks. The memory can store various objects or data, including caches, classes, frameworks, applications, modules, backup data, jobs, web pages, web page templates, data structures, database tables, repositories, and dynamic information. Types of objects and data stored in memory can include parameters, variables, algorithms, instructions, rules, constraints, and references. Additionally, the memory can include logs, policies, security or access data, and reporting files. The processor and the memory can be supplemented by, or incorporated in, special purpose logic circuitry.

Implementations of the subject matter described in the present disclosure can be implemented on a computer having a display device for providing interaction with a user. Types of display devices can include, for example, a cathode ray tube (CRT), a liquid crystal display (LCD), a light-emitting diode (LED), and a plasma monitor. Display devices can include a keyboard and pointing devices including, for example, a mouse, a trackball, or a trackpad. User input can also be provided to the computer through the use of a touchscreen, such as a tablet computer surface with pressure sensitivity or a multi-touch screen using capacitive or electric sensing. Other kinds of devices can be used to provide for interaction with a user, including to receive user feedback including, for example, sensory feedback including visual feedback, auditory feedback, or tactile feedback. Input from the user can be received in the form of acoustic, speech, or tactile input. In addition, a computer can interact with a user by sending documents to, and receiving documents from, a device that is used by the user. For example, the computer can send web pages to a web browser on a user's client device in response to requests received from the web browser.

The term “graphical user interface,” or “GUI,” can be used in the singular or the plural to describe one or more graphical user interfaces and each of the displays of a particular graphical user interface. Therefore, a GUI can represent any graphical user interface, including, but not limited to, a web browser, a touch screen, or a command line interface (CLI) that processes information and efficiently presents the information results to the user. In general, a GUI can include a plurality of user interface (UI) elements, some or all associated with a web browser, such as interactive fields, pull-down lists, and buttons. These and other UI elements can be related to or represent the functions of the web browser.

Implementations of the subject matter described in this specification can be implemented in a computing system that includes a back end component (for example, as a data server), or that includes a middleware component (for example, an application server). The components of the system can be interconnected by wireless digital data communication (or a combination of data communication) in a communication network. Examples of communication networks include a wireless local area network (WLAN) (for example, using 802.11 a/b/g/n or 802.20 or a combination of protocols), all or a portion of the Internet, or any other communication system or systems at one or more locations (or a combination of communication networks). The network can communicate with, for example, Internet Protocol (IP) packets, frame relay frames, asynchronous transfer mode (ATM) cells, voice, data, or a combination of communication types between network addresses.

The computing system can include clients and servers. A client and server can generally be remote from each other and can typically interact through a communication network. The relationship of client and server can arise by virtue of computer programs running on the respective computers and having a client-server relationship.

While this specification contains many specific implementation details, these should not be construed as limitations on the scope of what may be claimed, but rather as descriptions of features that may be specific to particular implementations. Certain features that are described in this specification in the context of separate implementations can also be implemented, in combination, in a single implementation. Conversely, various features that are described in the context of a single implementation can also be implemented in multiple implementations, separately, or in any suitable sub-combination. Moreover, although previously described features may be described as acting in certain combinations and even initially claimed as such, one or more features from a claimed combination can, in some cases, be excised from the combination, and the claimed combination may be directed to a sub-combination or variation of a sub-combination.

Particular implementations of the subject matter have been described. Other implementations, alterations, and permutations of the described implementations are within the scope of the following claims as will be apparent to those skilled in the art. While operations are depicted in the drawings or claims in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed (some operations may be considered optional), to achieve desirable results. In certain circumstances, multitasking or parallel processing (or a combination of multitasking and parallel processing) may be advantageous and performed as deemed appropriate.

Moreover, the separation or integration of various system modules and components in the previously described implementations should not be understood as requiring such separation or integration in all implementations, and it should be understood that the described program components and systems can generally be integrated together in a single software product or packaged into multiple software products.

Accordingly, the previously described example implementations do not define or constrain the present disclosure. Other changes, substitutions, and alterations are also possible without departing from the spirit and scope of the present disclosure.

Furthermore, any claimed implementation is considered to be applicable to at least a computer-implemented method; a non-transitory, computer-readable medium storing computer-readable instructions to perform the computer-implemented method; and a computer system comprising a computer memory interoperably coupled with a hardware processor configured to perform the computer-implemented method or the instructions stored on the non-transitory, computer-readable medium.

A number of embodiments of these systems and methods have been described. Nevertheless, it will be understood that various modifications may be made without departing from the spirit and scope of this disclosure.

Claims
1-20. (canceled)

21. A sensor system for monitoring a vascular access point of a subject, wherein the sensor system comprises:

a plurality of sensors configured to capture sensor data characterizing a physiological state of the subject in a vicinity of the vascular access point of the subject;
one or more computer processors; and
one or more storage devices communicatively coupled to the one or more computer processors, wherein the one or more storage devices store instructions that, when executed by the one or more computer processors, cause the one or more computer processors to perform operations comprising: adjusting operational parameters of the plurality of sensors using one or more machine learning models to optimize the operational parameters of the plurality of sensors with reference to an objective function; and after adjusting the operational parameters of the plurality of sensors using the one or more machine learning models: receiving sensor data captured by the plurality of sensors; and processing the sensor data captured by the plurality of sensors to classify a state of the vascular access point of the subject; and generating an alert based on the classification of the state of the vascular access point of the subject.
22. The sensor system of claim 21, wherein adjusting the operational parameters of the plurality of sensors using the one or more machine learning models comprises:

processing one or more model inputs, by the one or more machine learning models, to generate one or more machine learning model outputs that define an update to values of the operational parameters of the plurality of sensors; and
adjusting the values of the operational parameters of the plurality of sensors based on the one or more machine learning model outputs.
23. The system of claim 22, wherein adjusting the values of the operational parameters of the plurality of sensors based on the one or more machine learning model outputs comprises:

determining updated values of the operational parameters of the plurality of sensors based on the one or more machine learning model outputs; and
determining that the updated values of the operational parameters of the plurality of sensors are consistent with one or more constraints on the values of the operational parameters of the plurality of sensors.
24. The system of claim 21, wherein each of the one or more machine learning models is parameterized by a set of machine learning model parameters having trained values that have been determined by a machine learning training technique.

25. The system of claim 21, wherein for one or more of the plurality of sensors, adjusting the operational parameters of the sensor comprises:

adjusting a sampling rate of the sensor.
26. The system of claim 21, wherein for one or more of the plurality of sensors, adjusting the operational parameters of the sensor comprises:

adjusting an ultrafiltration parameter of the sensor.
27. The system of claim 21, wherein for one or more of the plurality of sensors, adjusting the operational parameters of the sensor comprises:

adjusting a calibration parameter of the sensor.
28. The system of claim 21, wherein the objective function is dependent on at least power usage by the plurality of sensors.

29. The system of claim 21, wherein the objective function is dependent on at least on signal-to-noise ratio (SNR) of sensor data generated by the plurality of sensors.

30. The system of claim 21, wherein the objective function is dependent on at least sensor performance of the plurality of sensors.

31. The system of claim 21, wherein the operations further comprise showing the alert on a display.

32. The system of claim 21, wherein the one or more machine learning models comprise one or more neural networks models.

33. The system of claim 21, wherein the sensor system is a wearable system.

34. The system of claim 21, wherein processing the sensor data captured by the plurality of sensors to classify the state of the vascular access point of the subject comprises conditioning the sensor data captured by the plurality of sensors, the conditioning comprising, for each of the plurality of sensors:

processing raw sensor data received from the sensor using a conditioning machine learning model corresponding to the sensor, in accordance with trained values of a set of conditioning machine learning model parameters, to generate a set of conditioning parameters; and
applying a conditioning operation, parameterized by the set of conditioning parameters generated by the conditioning machine learning model corresponding to the sensor, to the raw sensor data received from the sensor.
35. The system of claim 34, for each sensor of the plurality of sensors, the trained values of the set of conditioning machine learning model parameters corresponding to the sensor are subject-specific values that are determined through training, by a machine learning training technique, on subject-specific raw sensor data characterizing the physiological state of the subject.

36. The system of claim 21, wherein processing the sensor data captured by the plurality of sensors to classify the state of the vascular access point of the subject comprises:

determining differential readings between pairs of sensors of the plurality of sensors; and
determining whether the differential readings satisfy a threshold.
37. The system of claim 21, wherein classifying the state of the vascular access point of the subject comprises:

classifying whether the vascular access point is infected.
38. The system of claim 21, wherein classifying the state of the vascular access point of the subject comprises:

classifying an optimal transfer rate through the vascular access point.
39. A method of using a sensor system to monitor a vascular access point of a subject, wherein the sensor system comprises a plurality of sensors configured to capture sensor data characterizing a physiological state of the subject in a vicinity of the vascular access point of the subject, the method comprising:

adjusting operational parameters of the plurality of sensors using one or more machine learning models to optimize the operational parameters of the plurality of sensors with reference to an objective function; and after adjusting the operational parameters of the plurality of sensors using the one or more machine learning models: receiving sensor data captured by the plurality of sensors; and processing the sensor data captured by the plurality of sensors to classify a state of the vascular access point of the subject; and generating an alert based on the classification of the state of the vascular access point of the subject.
40. One or more non-transitory computer storage media storing instructions that when executed by one or more computers cause the one or more computers to perform operations for using a sensor system to monitor a vascular access point of a subject, wherein the sensor system comprises a plurality of sensors configured to capture sensor data characterizing a physiological state of the subject in a vicinity of the vascular access point of the subject, the operations comprising:

adjusting operational parameters of the plurality of sensors using one or more machine learning models to optimize the operational parameters of the plurality of sensors with reference to an objective function; and after adjusting the operational parameters of the plurality of sensors using the one or more machine learning models: receiving sensor data captured by the plurality of sensors; and processing the sensor data captured by the plurality of sensors to classify a state of the vascular access point of the subject; and generating an alert based on the classification of the state of the vascular access point of the subject.

Glad I could help. Be sure to read it all. - Forecasting Artificial Intelligence Trends in Health Care: Systematic International Patent Analysis in the Journal of Medical Internet Research AI. Study takes an interesting deep dive into healthcare AI patents from the last decade, analysing over 10,000 of them from 2012 to 2022. Through this exploration, shed light on the growth trends, the direction AI innovation is heading in different medical specialties, and the potential future implications of AI and Machine Learning (ML) in healthcare.

Our journey into the world of healthcare AI innovation and regulation started in 2019-2020 with a fair amount of research resulting in this study published in Nature digital medicine. A fascinating starting point for that venture was that although the FDA had been approving AI-based devices for years, they hadn’t initially distinguished them as a unique category. We decided to sift through these approvals and identify those that were AI-based. This resulted in the creation of an open-access database, which we shared with the FDA. Imagine our delight when, a year later, the FDA published their own database and cited a source!

To illustrate this work, we also created an infographic, “The Current State of FDA-Approved, AI-Based Medical Devices,” which, to immense pleasure, has been downloaded millions of times and has been updated regularly since then.

Must read - Wearable health monitoring system
Devices, systems and methods are disclosed which relate to remotely monitoring the health of an individual. The individual wears a health monitoring device, with an attached strap, capable of sensing characteristics of the individual. These characteristics may include voice level and tone, movements, blood pressure, temperature, etc. The device allows individuals to constantly monitor their health without having to physically visit a doctor or other health care professional. Wireless communication, for instance with an Internet Protocol Television (IPTV) set-top box, allows measurements to be made and evaluated by a ‘computerized’ healthcare service provider. For a more accurate evaluation, measurements are sent over the INTERNET to a service. The device communicates with services in order to diagnose the individual based upon the characteristics.

BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to monitoring devices. More specifically, the present invention relates to a wearable health monitoring device for remotely diagnosing patients.
2. Background of the Invention
Visiting a doctor is often important to ensure the health of an individual, not only when the individual is sick, but also for regular check-ups. However, doctor visits can be costly, time consuming, and sometimes unpleasant. Many illnesses require constant visits to the doctor for monitoring. This monitoring may simply be monitoring an individual's blood pressure, determining if any changes have occurred, etc. Unfortunately, to receive this monitoring, the individual may have to drive a long distance, sit in a waiting room, see the doctor, and then drive the long distance home. This is not ideal, as it may end up taking most of, if not all of, the day.
Today there are many health statistics that are important to diagnosing the average individual. Besides statistics like pulse and temperature, muscular flexibility, hand-eye coordination, and basic reflexes are not only useful in everyday circumstances, but are indications of general health. However, since the enactment of The Health Insurance Portability and Accountability Act (HIPAA), certain precautions must be taken to keep this information between the patient and his or her doctor.
When visiting the doctor, people often complain about past conditions or episodes. However, a doctor can only test the patient's current status and ask them questions to recall how they felt during the past episode. Patient accounts can be uninformative and unreliable. Patients largely do not recall things such as instant pulse, blood pressure, temperature, etc. For instance, a patient may remember feeling cold, which can indicate a high temperature, but there is no way for the doctor to determine the exact temperature or even if the patient had a fever at all.
Many individuals would much rather stay in the privacy of their own home. However, the cost of frequent house calls by a doctor or other health care professional is too much for most individuals. Thus, staying at home is currently not a real option.
IPTV is a system through which digital television service is delivered using the architecture and networking methods of Internet protocols over a packet-switched network infrastructure, such as the Internet and broadband Internet access networks, instead of being delivered through traditional radio frequency broadcast and cable television formats.
What is needed is a way to monitor an individual's health while the individual is at home or away from the doctor's office.
SUMMARY OF THE INVENTION
The present invention provides devices and methods for remotely monitoring the health of an individual. The individual wears a health monitoring device, with an attached strap, capable of sensing characteristics of the individual. These characteristics may include voice level and tone, movements, blood pressure, temperature, etc. The device allows the individual to constantly monitor his or her health without having to physically visit a doctor or other health care professional. Wireless communication, for instance with an Internet Protocol Television (IPTV) set-top box allows measurements to be made and evaluated by a ‘computerized’ healthcare service provider. For a more accurate evaluation, measurements are sent over the INTERNET to a service. The device communicates with services in order to diagnose the individual based upon the characteristics.
Embodiments of the present invention work with an IPTV-based application where an individual uses a health monitoring device, in the form of a wearable wireless voice remote, to interact with a ‘computerized’ healthcare service provider at his or her home. The individual interacts with the service using their voice and body movements, such as touching the nose or the toe within a defined time window, according to the video instructions showing on the IPTV screen. The invention collects data based upon these interactions as well as inputs of the individual's health statistics, such as pulse, temperature, etc. in order for a live or virtual health care professional to diagnose the individual.
In an exemplary embodiment of the present invention, the invention is a device for monitoring health. The device includes a processor, a memory in communication with the processor, a remote health monitor logic on the memory, a health profile database on the memory, a wireless transceiver in communication with the processor, a housing enclosing the processor, the memory, and the wireless transceiver, a microphone in communication with the processor, a speaker in communication with the processor, and a strap coupled with the housing. A set-top box receives a voice sample and detects a distance of the wireless transceiver from the set-top box to monitor a user's health by producing an audio tone from a speaker; sending a data packet from the set-top box at the same time the audio tone is produced; and calculating the distance using the difference in time between receipt of the audio tone by the microphone and the data packet by the wireless transceiver.
In another exemplary embodiment of the present invention, the invention is a system for monitoring health. The system includes a wireless health monitoring device, a set-top box having a box memory in communication with the wireless health monitoring remote, and a box health monitor logic on the box memory. The set-top box receives a voice sample and detects a distance of the wireless transceiver from the set-top box to monitor a user's health by producing an audio tone from a speaker; sending a data packet from the set-top box at the same time the audio tone is produced; and calculating the distance using the difference in time between receipt of the audio tone by the microphone and the data packet by the wireless transceiver.
In a further exemplary embodiment of the present invention, the invention is a method for monitoring health. The method includes connecting a health monitoring device to a set-top box, receiving a health sample from the health monitoring device, receiving a voice sample from the health monitoring device, evaluating the voice sample with a voice baseline, instructing a user to perform a physical move, receiving a position sample determinative of the physical move, and evaluating the position with a position baseline using time difference of arrival between an audio tone and a data packet.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 shows a system for monitoring health, according to an exemplary embodiment of the present invention.
FIG. 2 shows a user interacting with a health monitoring device, according to an exemplary embodiment of the present invention.
FIGS. 3A and 3B show a health monitoring device and its components, according to an exemplary embodiment of the present invention.
FIGS. 4A, 4B, and 4C show a user performing tasks with a health monitoring device according to instructions, according to an exemplary embodiment of the present invention.
FIG. 5 shows a network-based system for monitoring health, according to an exemplary embodiment of the present invention.
FIG. 6 shows a method of remotely monitoring health, according to an exemplary embodiment of the present invention.
FIG. 7 shows an IPTV set-top box and a television listing instructions, according to an exemplary embodiment of the present invention.
DETAILED DESCRIPTION OF THE INVENTION
The present invention provides devices and methods for remotely monitoring the health of an individual. The individual wears a health monitoring device, with an attached strap, capable of sensing characteristics of the individual. These characteristics may include voice level and tone, movements, blood pressure, temperature, etc. The device allows the individual to constantly monitor his or her health without having to physically visit a doctor or other health care professional. Wireless communication, for instance with an Internet Protocol Television (IPTV) set-top box allows measurements to be made and evaluated by a ‘computerized’ healthcare service provider. For a more accurate evaluation, measurements are sent over the INTERNET to a service. The device communicates with services in order to diagnose the individual based upon the characteristics.
Embodiments of the present invention work with an IPTV-based application where an individual uses a health monitoring device, in the form of a wearable wireless voice remote, to interact with a ‘computerized’ healthcare service provider at his or her home. The individual interacts with the service using his or her voice and body movements, such as touching the nose or the toe within a defined time window, according to the video instructions showing on the IPTV screen. The invention collects data based upon these interactions as well as inputs of the individual's health statistics, such as pulse, temperature, etc. in order for a live or virtual health care professional to diagnose the individual.
In IPTV-based embodiments, an IPTV-based video-oriented service is presented to the patient either through a pre-recorded video of a human medical service provider, such as an orthopedic surgeon, or an interactive video session with a computer-animation based avatar. In embodiments using an avatar, the patient follows the instructions provided by the ‘virtual doctor’ on the IPTV screen to report and/or show his or her health status through speech, body movement, walking, stretching, etc.
With a health monitoring device and a pre-paid (or pay-per-visit) healthcare plan offered by the IPTV-based service provider, the patient completes a routine check-up at his or her home by interacting with a virtual doctor over IPTV. The patient responds to the video instructions shown on the IPTV screen through physical responses in real time such as using his or her voice (vocal level, intelligibility, clarity, recall/repeat the words heard from the videos and/or from memory) and body movements (attempt a physical reach in responding to an instruction like ‘please press and hold Talk button and touch your left toe’). The system records the patient's spatial-temporal movement via distance and direction measurements between the health monitoring device and a reference position such as the TV or the transceiver connected to the IPTV set-top box, the speed with which the patient moves, visual/cognitive alertness through answering the questions imposed by the virtual doctor on the IPTV screen, motor skills, etc. For example, the virtual doctor may ask the patient to report the objects displayed on the TV via voice response in a given time window in order to determine alertness. Motor skill may be determined by following the instructions given by the virtual doctor on the IPTV screen, such as “please press Talk button three times while you walk towards me (i.e., towards the TV)” or “please say ‘I am ok’ after you slowly bend your body forward towards the floor and repeat this 5 times”.
For example, the user may complain of an injured shoulder. The doctor or virtual doctor, through the IPTV set-top box, may tell the user to raise his or her hand in order to test the user's range of motion. The user holds the health monitoring device in his or her hand and moves the arm as instructed. The distance and direction from the health monitoring device are detected by the IPTV set-top box. These distance and directional readings are processed by the processor according to a box health monitor logic employing algorithms of the IPTV set-top box, herein explained below, to determine the movements that were made by the user. These readings may be used alone or compared with baseline readings for the user in order to determine whether the user has a limited range of motion. If so, the limited range of motion becomes a factor in the diagnosis of the user.
In an exemplary embodiment when the TV and the set-top box are placed close together, or a designated speaker is placed close to the set-top box, the distance between the health monitoring device and, for instance, the TV, can be determined by measuring the time difference of arrival (TDOA) between two signals, both of which are issued by the set-top box. For TV embodiments, a user may select to use an internal speaker or an external speaker in communication with the TV as the designated speaker. For speaker embodiments, the designated speaker can be a small dedicated speaker coupled to or within the set-top box, or one in a set of external speakers such as in a surround sound system. In either embodiment, the user selects the speaker closest to the set-top box to use as the designated speaker. The first signal is a data packet sent by the IPTV set-top box to the health monitoring device over their wireless data communication channel. The second signal is a specific audio tone played by the designated speaker as commanded by the IPTV set-top box. The timing of issuance of the set-top box commands which result in the two signals should compensate for various delays such as the command processing delay and designated speaker response delay so that both signals begin simultaneously. Because the data packet is transmitted over RF, the health monitoring device shall receive the packet prior to detecting the audio tone. The difference in time between receiving the packet and detecting the tone is used to calculate the distance between the designated speaker and the health monitoring device. Denoting the distance to be calculated as D, the propagation speed of audio tone as Va and the propagation speed of radio as Vr, the measured time difference as T, their relationship can be expressed as:
D/V a -D/V r =T
or further:
D=TV a V r/(V r -V a).
Alternately, a designated speaker may not be near the set-top box. Geometry can be used to correct for the difference in distance between the set-top box and the health monitoring device, and the designated speaker and the health monitoring device. When there are multiple audio speakers in the TV room, a designated speaker may not be selected. The TDOA method described above is used to find the distances between the health monitoring device and each of the speakers. Combining these distances with the known distances among the speakers themselves, the direction from the TV to the health monitoring device can be computed with simple geometry.
When the set-top box and the TV are not co-located, the time difference between the two signals can be used in motion/movement related diagnostics. For instance, the time difference between the two signals can be used to measure a change in relative position. Baseline TDOA measurements can be taken while the patient is in a normal condition and future diagnostic TDOA measurements can be compared with this baseline profile.
Various modes of using the present invention are also possible. In a network mode, the health monitoring device and/or the IPTV set-top box is connected to a network in communication with a live or virtual health care professional. In a local mode, the health monitoring device communicates with the IPTV set-top box to record various measurements either for a diagnosis based upon a database on the IPTV set-top box or for later sending to a network. A non-connected mode allows a user to carry the health monitoring device with them to record measurements throughout the day. These measurements are stored on a memory of the health monitoring device and are later sent to the IPTV set-top box or network when these are in communication with the health monitoring device.
Embodiments of the invention allow users to input various demographic categories they belong to as well as other symptoms they are having. These inputs provide the live or virtual healthcare professional information that may be important in diagnosing the user. For instance, the user may enter his or her age, height, and weight as well as flu-like symptoms he or she is having.
For the following description, it can be assumed that most correspondingly labeled structures across the figures (e.g., 132 and 232, etc.) possess the same characteristics and are subject to the same structure and function. If there is a difference between correspondingly labeled elements that is not pointed out, and this difference results in a non-corresponding structure or function of an element for a particular embodiment, then that conflicting description given for that particular embodiment shall govern.
FIG. 1 shows a system for monitoring health, according to an exemplary embodiment of the present invention. This system may be used at the local level, or may be connected to a network. In this embodiment, the system includes a health monitoring device 100, a set-top box 120, and a television 130. Health monitoring device 100 includes a push button 104, a microphone 102, a speaker 106, and a health profile database 118 on a memory. Health monitoring device 100 is a small wireless device capable of detecting the user's voice, movements, health statistics, etc. Health monitoring device 100 attaches to a user, such as with an attached strap. Health monitoring device 100 may take an ergonomic form such that it fits comfortably in the user's hand. Health monitoring device 100 uses various sensors to take readings on the user as well as the user's environment. These reading may include the user's temperature, the room temperature, the user's pulse, the oxygen level in the user's blood, the user's blood sugar level, the user's blood alcohol level, the user's pupil dilation, the user's pupil movement, the user's range of motion, the user's speed of motion, as well as other readings generally taken at a doctor's office. The sensors used may be integrated into health monitoring device 100 or may be coupled to health monitoring device 100 using adaptors. A wireless transceiver within health monitoring device 100 communicates with a transceiver 122 on set-top box 120 to detect motion of the user, receive commands from set-top box 120, send data to set-top box 120, etc. A distance 132 between the wireless transceiver and transceiver 122 is determined. The wireless transceiver in health monitoring device 100 may use Radio Frequency (RF) communication technologies such as cellular, wireless personal area networks, e.g., BLUETOOTH wireless networks, wireless local area networks, e.g., WI-FI wireless networks, low-rate wireless personal area networks, e.g., ZIGBEE wireless networks, radiofrequency identification (RFID), Ultra Wide Band (UWB), etc., or optical communication technologies such as infrared (IR), laser, Visible Light Communication (VLC), etc. Push button 104 on health monitoring device 100 allows the user to respond to commands from or initiate commands to set-top box 120 or health monitoring device 100. Push button 104 may be used to signify the start of movements, can turn microphone 102 on or off, etc. In other embodiments, push button 104 is a touch-sensitive biometric multifunction sensor. Push button 104 may also serve to authenticate the user, as in some embodiments push button 104 includes a fingerprint scanner or other biometric reader or optical scanner. Other authentication methods may be through voice authentication when the user presses push button 104, iris authentication, etc. Authentication is useful for health monitoring device 100, as the monitoring of the user may be compared with previous sessions, the diagnosis of the individual may be private, etc., and the identity of the individual is important. Additionally, push button 104 may detect the user's pulse, dexterity, etc. Speaker 106 gives the user commands, alerts the user, etc. For instance, in a non-connected mode, the user may not be near IPTV set-top box 120 or television 130 and must otherwise receive commands. Microphone 102 provides an input for speech or sounds from the user. This may be in response to commands from television 130 or set-top box 120, to select functions, to call for help, etc. Health profile database 118 stores inputs, vital signs, etc. from the user. Health profile database 118 also contains baseline values for users of the device. These baseline values may be set by the user at an initial diagnosis, when setting up the system, etc. or may be set at the office of a doctor or healthcare professional.
Set-top box 120 includes a remote health monitoring logic 124 on a memory and a transceiver 122. Remote health monitoring logic 124 includes instructions for set-top box 120 to interact with health monitoring device 100. These instructions are carried out by a processor on set-top box 120 which commands other components of set-top box 120. Transceiver 122 may be a USB-based wireless local area network, e.g., a USB-based WI-FI wireless network, Access Point in communication with IPTV set-top box 120, an integrated transceiver, etc. Transceiver 122 allows set-top box 120 to connect with a network. This allows set-top box 120 to communicate with application servers, allowing the user to communicate with real and/or virtual 12 doctors, etc. Transceiver 122 also communicates with the wireless transceiver of health monitoring device 100 to receive inputs from health monitoring device 100, send outputs to health monitoring device 100, etc. Set-top box 120 communicates with television 130 using a wireless or wired connection. Television 130 displays visual output from set-top box 120. This may be instructions for the user, the user's results, diagnoses, etc.
In embodiments of the present invention, a protocol for setting baseline values enables the user to set baseline values at the doctor's office without the health monitoring device being at the office. This may be accomplished by using a similar health monitoring device to record baseline values or by using a plurality of other devices, such as a blood pressure cuff, a thermometer, etc., to set these values. These baseline values may be remotely set, with the office or service provider allowing the user to download the values or sync the values with the health monitoring device. In this way, all the patient records are at the doctor's office, but certain values like these baseline values may be sent to the device.
In embodiments of the present invention, the health monitoring device may take the form of earrings or may be embedded under the skin of the user. With earrings, the position and orientation of the user's head may easily be determined using wireless signals between wireless transceivers in the earrings and a transceiver on the IPTV set-top box. When embedded under the skin, the orientation and location of the embedded device is determined. By embedding the health monitoring device under the skin, other health statistics may also be taken.
In further embodiments of the present invention, the remote health monitoring logic may be stored on the memory of the health monitoring device and/or on a network in communication with the IPTV set-top box. By having the remote health monitoring logic in multiple locations, the different modes of the invention can be fully used.
Embodiments of the present invention use one or more adaptors to attach various measurement devices. The health monitoring device may include a finger clip for measuring the blood pressure of the user or may include an adaptor to attach such a finger clip. Similarly, the health monitoring device may include an adaptor for a stethoscope or other device.
FIG. 2 shows a user 240 interacting with a health monitoring device 200, according to an exemplary embodiment of the present invention. In this embodiment, health monitoring device 200 is used to measure user 240's speech. Users complaining of sore throats, coughing, and/or other illnesses or diseases often have discernable differences in their speech. Determining these differences may assist in the diagnosis of these users. In order to determine whether user 240's voice is different, user 240 first must record a baseline recording. This may be done during the initial setup of the system and health monitoring device 200, during the onset of symptoms, at an initial checkup, etc. and may be done by user 240 at home, with the assistance of a doctor at the doctor's office, etc. In order to create a baseline recording, user 240 speaks into microphone 202 of health monitoring device 200. Health monitoring device 200, an IPTV set-top box connected to a television, or a medical professional may instruct user 240 to say a word, phrase, etc. into microphone 202. When health monitoring device 200 instructs user 240, instructions are delivered through speaker 206. Health monitoring device 200 records user 240's speech 242 to a memory on health monitoring device 200, a memory on the IPTV set-top box, a memory on a network, etc. If the recorded speech has not been made or used before, the recorded speech is set as a baseline recording. If a baseline recording has already been created, the recorded speech is compared with the baseline recording. Differences in patterns, volume, tone, inflection, etc. are determined to assist with the diagnosis of the patient. The comparison may be done by a processor on health monitoring device 200, by a processor on the IPTV set-top box, by a processor on the network, etc.
FIGS. 3A and 38 show a health monitoring device 300 and its components, according to an exemplary embodiment of the present invention. Health monitoring device 300 may be used alone in a non-connected mode, may be used in conjunction with an IPTV set-top box in a local mode, or may be used in a network mode with a live or virtual healthcare professional. In this embodiment, health monitoring device 300 includes a microphone 302, a speaker 306, a push button 304, a strap 308, a wireless transceiver 315, a central processing unit 312, and a memory 314. Strap 308 wraps around the neck of the user. Strap 308 is a stretchable material, such as a coiled plastic ring resembling a common telephone cord. Strap 308 may be stretched in order for the user to move with health monitoring device 300 and may retract such that it is near the user's mouth when the user is speaking into microphone 302. Speaker 306 allows the user to receive instructions, listen to diagnoses, listen to music or a game, etc. Push button 304 provides an input and/or a switching mechanism as well as an input for authentication. Push button 304 may be used in order to switch microphone 302 on and off. Additionally, push button 304 may contain sensors in order to monitor the user's pulse, the user's temperature, the room temperature, etc. Microphone 302 provides an input for health monitoring device. Microphone 302 detects voice volume, tone, inflection, etc. and sends this data to CPU 312. These may be compared with baseline values previously stored in health monitoring device 300 or by an IPTV set-top box or server on a network. When comparing with baseline values on health monitoring device 300, CPU 312 runs a remote health monitor logic 316 on memory 314 that compares the values recorded by microphone 302 with values stored on a health profile database 318 on memory 314. CPU 312 commands components of health monitoring device 300 according to remote health monitor logic 316 as well as other logic on memory 314. Memory 314 stores logic, data, etc. Together, CPU 312 and memory 314 may serve to buffer speech and or movement data to ensure the data is accurate. Memory 314 also stores device identification in order to authenticate health monitoring device 300 with an IPTV set-top box and/or network. Health profile database 318 stores data concerning the user or users. This data may include a personal profile including height, weight, body fat, medical history, etc. as well as data received from user inputs. These user inputs may be in response to instructions from a doctor, a virtual doctor, etc. Wireless transceiver 315 allows health monitoring device 300 to wirelessly communicate with a network, other wireless devices such as an IPTV set-top box, etc. Wireless transceiver 315 may use Radio Frequency (RF) communication technologies such as cellular, wireless personal area networks, e.g., BLUETOOTH wireless networks, wireless local area networks, e.g., WI-FI wireless networks, low-rate wireless personal area networks, e.g., ZIGBEE wireless networks, radio-frequency identification (RFID), Ultra Wide Band (UWB), etc., or optical communication technologies such as infrared (IR), laser, Visible Light Communication (VLC), etc.
Embodiments of the present invention also include an accelerometer 313 within health monitoring device 300. Accelerometer 313 determines the orientation of health monitoring device 300. Accelerometer 313 is also used to detect motions of health monitoring device 300, such as for performing requested motions, playing games, etc. In embodiments of the present invention, accelerometer 313 is sensitive such that it may also be used to determine the pulse of the user, as the pulse creates small movements. Readings from accelerometer 313 are sent to CPU 312 to assist health monitoring device 300 in determining the extent of motions by the user.
FIGS. 4A, 4B, and 4C show a user 440 performing tasks with a health monitoring device 400 according to instructions, according to an exemplary embodiment of the present invention. These instructions may be delivered by health monitoring device 400, by an IPTV set-top box connected to a television, by a live or virtual medical professional, etc. In each of these embodiments, user 440 is wearing a strap 408 around user 440's neck. Strap 408 is a stretchable material such that user 440 can hold health monitoring device 400 in his or her hand and perform motions. When health monitoring device 400 is not in user 440's hand, health monitoring device 400 is closer to user 440's mouth and ears such that user 440 can receive instructions, make voice commands, record speech, etc.
FIG. 4A shows a user 440 recording speech on a health monitoring device 400, according to an exemplary embodiment of the present invention. In this embodiment, a strap 408 is around user 440's neck. Health monitoring device 400 is in a position to instruct user 440 and record user 440's speech. Health monitoring device 400 is hanging, with strap 408, in an unstretched position. This position is recorded by the health monitor logic and serves as a base position from which the movement in the next Figures is measured.
FIG. 4B shows a user 440 performing arm motions with a health monitoring device 400, according to an exemplary embodiment of the present invention. In this embodiment, a strap 408 is around user 440's neck and user 440 is extending an arm out in response to a command from health monitoring device 400 or an IPTV set-top box through a television. This embodiment is useful, for instance, when measuring user 440's range of motion. The distance moved by health monitoring device 400, the speed of movement, angles of movement, etc. are recorded by health monitoring device as data. This data may be compared with baseline data from the position in FIG. 4A to determine if user 440's movement is limited. When a baseline is taken right after an injury, it can be determined from the data whether user 440's injury is improving. When a baseline is taken when user 440 is healthy, the data can determine limitations in user 440's motion. The data is determined by measurements made by components of health monitoring device 400 and/or the IPTV set-top box. These measurements may use changes in distance and direction from the health monitoring device to a reference position such as the TV using accelerometer measurements, etc. in order to determine user 440's movement. Measurements are sent through a processor of health monitoring device 400 and/or the IPTV set-top box to determine the movements and compare the movements with the baseline data.
FIG. 4C shows a user 440 wearing a health monitoring device 400 and bending over at the waist, according to an exemplary embodiment of the present invention. In this embodiment, user 440 has a strap 408 connected to health monitoring device 400 around user 440's neck. User 440 has been instructed to touch his or her toes. With strap 408 around user 440's neck, user 440 slowly bends at the waist to touch the toes. The location of health monitoring device 400 moves as user 440 bends lower. Measurements of the distance bent down as well as other measurements are taken by health monitoring device 400 using infrared, wireless local area networks, e.g., WI-FI wireless networks, TDOA, accelerometers, etc. similar to the measurements in FIG. 48. These measurements may be compared to a baseline measurement, may form the baseline measurement, etc.
FIG. 5 shows a network-based system for monitoring health, according to an exemplary embodiment of the present invention. In this embodiment, a health monitoring device 500 and/or an IPTV set-top box 520 are in communication with a server 552 connected through Internet 550. This allows the user to communicate and send data back and forth with a live or virtual health care professional. When taking voice measurements, voice measurements recorded by a microphone 502 of health monitoring device 500 are sent through a wireless transceiver of health monitoring device 500 to a transceiver 522 of IPTV set-top box 520. Transceiver 522 forwards these voice measurements over Internet 550 to server 552 where they may be processed by a processor based upon a remote health monitor logic 554 on a memory of server 552. Measurements taken by push button 504 may be similarly forwarded to server 552. Server 552 may be an IPTV healthcare service node with automatic speech recognition (ASR) with speaker adaptation capability. This allows server 552 to recognize speech from the user. When sending commands to the user, server 552 sends commands over Internet 550 to transceiver 522 of IPTV set-top box 520. These commands may be selected from a database on server 552 or network 550 based upon remote health monitor logic 554. A health care professional may upload these commands as well as diagnoses to the database of server 552. The commands are either delivered through television 530 connected to IPTV set-top box 520 or are forwarded through transceiver 522 to health monitoring device 500 where they are delivered through a speaker 502.
In embodiments of the present invention using a live doctor or other healthcare professional, the doctor can similarly connect to the server using an IPTV set-top box, video conferencing equipment, etc. In this way, the doctor communicates directly with the user over the Internet. In embodiments utilizing a virtual healthcare professional, a memory, a processor, and the remote health monitor logic on the server store and determine responses or diagnoses based upon inputs by the user.
In embodiments of the present invention, the system may use games and/or videos in order to test user responses or to calibrate the health monitoring device. These games and/or videos may be stored on the server, on the IPTV set-top box, etc. Using the health monitoring device for games allows the system to determine, for instance, the ability to follow commands, testing cognitive functioning, etc. Games may also serve to test the user's flexibility, reaction time, range of motion, etc. Games or videos may also be used when calibrating the health monitoring device or setting a baseline. Games may be used to establish certain motions within the system for calibration. Games or videos may be used, for instance, to raise the user's heart rate to a certain level necessary for calibration or a baseline.
FIG. 6 shows a method of remotely monitoring health, according to an exemplary embodiment of the present invention. In this embodiment, the method begins by calibrating the health monitoring device S660. The calibration may be accomplished by setting baseline levels for speech, motions, positions, etc. With the health monitoring device calibrated, the health monitoring device or an IPTV set-top box through a television requests a voice sample from the user S661. The user speaks into the microphone of the health monitoring device which records the voice sample. The health monitoring device or IPTV set-top box then requests a physical sample S662. For example, the user may be requested to bend at the waist as far as possible. A transceiver on the IPTV set-top box communicates with a wireless transceiver of the health monitoring device to create data for the physical sample. With the voice sample and/or the physical sample, a processor, according to logic on a memory, analyzes the data and compares the voice sample and/or physical sample with baseline samples S663. The analyzing includes compiling the inputs to determine characteristics of the voice sample and/or physical sample. The comparison determines whether the voice sample and/or physical sample are different than the baseline samples and, if so, how they are different. These differences help to influence the diagnosis. The processor determines whether the results are irregular S664. A result is irregular if the voice sample and/or physical sample is different from the baseline sample by a set amount. This may be an amount set by the healthcare professional based upon the test being performed. If the results are irregular, the user is alerted S665. If the results are not irregular but the user is not appropriately responding, the user or the processor may determine that the user is in critical condition S666. This may be if the user is not showing a pulse, not responding to prompts, etc. The user or processor may also determine that the user is in critical condition S666 after the user has been alerted S665. If the user or processor does not signal that the user is in critical condition, the method ends. If the user is in critical condition, the health monitoring device, the IPTV set-top box, or a server on the network alerts authorities S667. The user may be given a certain window of time to respond before notifying authorities. For instance, the system may automatically call an ambulance to the location of the user if the user has been prompted and does not respond within five minutes.
FIG. 7 shows an IPTV set-top box 720 and a television 730 listing instructions, according to an exemplary embodiment of the present invention. In this embodiment, IPTV set-top box 720 communicates with a health monitoring device to determine whether the user is responding to the instructions. The instructions are viewed on the screen of television 730. Each instruction is listed with a corresponding check box that is marked when the instruction is completed by the user with the health monitoring device. For instance, in the figure, the first instruction is to connect the health monitoring device 770. The user has already connected the health monitoring device to the IPTV set-top box. Thus, check box 772 has been marked.
All of the aforementioned combinations should be customizable to suit the user. In some cases it may even be advantageous to provide input models suited to various disabilities, thus improving the usefulness of the device for the largest possible user base. Beyond initial settings, this mechanism should be automatic, autonomous and much more user friendly than the alternatives.
The foregoing disclosure of the exemplary embodiments of the present invention has been presented for purposes of illustration and description. It is not intended to be exhaustive or to limit the invention to the precise forms disclosed. Many variations and modifications of the embodiments described herein will be apparent to one of ordinary skill in the art in light of the above disclosure. The scope of the invention is to be defined only by the claims appended hereto, and by their equivalents.
Further, in describing representative embodiments of the present invention, the specification may have presented the method and/or process of the present invention as a particular sequence of steps. However, to the extent that the method or process does not rely on the particular order of steps set forth herein, the method or process should not be limited to the particular sequence of steps described. As one of ordinary skill in the art would appreciate, other sequences of steps may be possible. Therefore, the particular order of the steps set forth in the specification should not be construed as limitations on the claims. In addition, the claims directed to the method and/or process of the present invention should not be limited to the performance of their steps in the order written, and one skilled in the art can readily appreciate that the sequences may be varied and still remain within the spirit and scope of the present invention.

In recent years there has been an explosion in the field of healthcare wearables; high-tech devices now range from patches, rings, watches and scarfs to insoles, headbands, contact lenses and glasses. Their functionality encompasses everything from counting steps to diagnosing Alzheimer’s disease, tracking fertility to delivering wound sealant in the event of trauma. The wearable medical device market is expected to reach over US $74 billion by 2028 and the COVID crisis has only driven innovation in this area1.

Patenting this emerging technology is of paramount importance to many innovator companies. The devices themselves are patentable at the European Patent Office (EPO), provided the claims meet the usual patentability requirements including novelty and inventive step. However, claims to the use of these devices can fall foul of exclusions to patentability enshrined in the European Patent Convention (EPC). In particular, as healthcare wearables become increasingly sophisticated, use claims associated with such devices are more likely to contravene EPC provisions which prevent the patenting of methods of treatment by surgery and therapy, and methods of diagnosis practised on the human or animal body.

But that does not, by any means, preclude the allowance of all such claims. The way in which these exclusions are interpreted by the EPO can leave useful windows of patentability. It can also lead to the refusal of claims that would, at first glance, not appear to be excluded.


1. Methods of diagnosis are excluded from patentability at the EPO

Article 53(c) EPC recites that:
“European patents shall not be granted in respect of…diagnostic methods practised on the human or animal body”.

Therefore, while the devices themselves may be patentable, any claims to the use of such a device in a diagnostic method may be excluded from patentability. However, the relatively narrow way in which “diagnostic methods” are defined by the EPO can leave room for manoeuvre, as discussed below.


2. Method claims will not be excluded if any of the four “necessary phases” are missing

To be excluded from patentability as a method of diagnosis, a claim must include method steps relating to all of the following phases:

(i) the examination phase, involving the collection of data;
(ii) the comparison of these data with standard values;
(iii) the finding of any significant deviation (i.e. a symptom) during the comparison; and
(iv) the attribution of the deviation to a particular clinical picture (i.e. the deductive medical or veterinary decision phase2).

If any of these steps are missing, the method will not be excluded under Article 53(c) EPC.

Example: Claims to the use of smart contact lenses that diagnose diabetes by measuring glucose concentration in tears might avoid the exclusion by claiming only the use of biosensors within the smart contact lenses to measure glucose concentration (step i), without claiming the subsequent analysis of the data and the diagnosis of diabetes (steps (ii) - (iv)).

Tips: If planning to avoid the exclusion by leaving one or more of steps (i) – (iv) out of the claim, it is important that the application is drafted in such a way that these steps are clearly not essential to the definition of the invention (i.e. that the invention is fully defined without these steps). If the claimed step(s) are only ever disclosed as part of a larger method of diagnosis, the EPO are likely to consider that the claimed step(s) cannot be isolated from this context so as to become the principle object of the invention3. In the above example, the invention could be framed as a method of data collection, and the description drafted to provide clear and unambiguous basis to claim this method without steps (ii) - (iv).

Furthermore, while the description can discuss the “missing steps”, for example the potential use of the claimed data collection method in a method of diagnosis (i.e. steps (ii) - (iv)), it is important not to include these steps in the dependent claims. Doing so can lead to an objection to the independent claim upon which these claims depend4.

One should also be aware that certain steps may be considered to be implicit. For example, steps relating to the comparison of data with standard values (step (ii)) may be considered to imply the finding of a significant deviation (step (iii))2,5.


3. Method claims may not be excluded if steps (i) to (iii) result in only an “intermediate finding” on the road to a clinical diagnosis

To be excluded from patentability, the method claim must include the final medical or veterinary decision phase, in other words, the diagnosis of “a particular clinical picture”2 (step (iv)).

Therefore, a claim relating to a method that stops short of actually diagnosing a particular clinical picture may not be excluded from patentability. Indeed, in one of the key decisions in the field of diagnostic methods, G1/04, the Enlarged Board of Appeal (EBA) confirmed that “intermediate findings” of diagnostic relevance must not be confounded with diagnosis for curative purposes and that a method for obtaining such findings did not constitute sufficient basis for denying patentability6.

Example: Claims to the use of sensors within smartphones and watches to detect gait speed, symmetry and stride length, which can be used by doctors to help diagnose Alzheimer’s disease, are likely to be patentable provided the step of diagnosing Alzheimer’s disease, or a medical condition which is indicative of Alzheimer’s disease, does not fall within the claimed steps of the method.

Tips: Take care when drafting the patent application as, if the diagnosis step would seem to be essential for the definition of the invention, simply leaving it out of the claim will lead to an objection from the EPO under Article 84 EPC on the grounds that an essential feature is missing2. This might arise if it is clear from the patent application as a whole that the results of the method would inevitably allow the attribution of the detected deviation to a particular clinical picture, even if that particular step is not claimed7,8.

It is important to note that the use of the language “attribution of the deviation to a particular clinical picture” in step (iv) does not imply that a specific disease needs to be diagnosed. The diagnosis may be of a broader nature than that and include “the determination of the nature of a medical or veterinary medicinal condition intended to identify or uncover a pathology”2... For example, in T1016/10, the detection of an “amyloidogenic disorder” was considered to represent a ”particular clinical picture”, even though amyloidogenic disorders include a wide range of different diseases and the claimed method did not allow the diagnosis of a specific one of these diseases9.


4. Method claims may not be excluded if the diagnosis is not “clinical” in nature

These days, healthcare wearables can detect and identify a wide range of physiological states beyond those directly related to diseases and medical conditions. As the Art.53(c) EPC exclusion applies only to the diagnosis of a particular clinical picture, such methods may be patentable.

Example: Claims to the use of healthcare wearables that monitor a women’s fertility for the purposes of aiding conception, by tracking physiological signals that act as markers for fluctuating hormone levels, are likely to be patentable.

Tips: Take care to draft the method claims using suitably specific language that does not include within its scope the diagnosis of diseases and medical conditions. For instance, considering the above example, monitoring a women’s fertility could theoretically be used for the diagnosis of a particular clinical picture.


5. Method claims may not be excluded if one of steps (i) to (iv) is a) technical and b) not carried out on the body

To be excluded under Article 53(c) EPC, all of the steps of a technical nature which are necessary for reaching the diagnosis must be carried out on the human or animal body2.

Step (iv) is always considered non-technical, but the Art.53(c) EPC exclusion can be avoided if at least: (i) the collection of data; (ii) the comparison of the data with standard values; or (iii) the finding of a significant deviation (i.e. a symptom) is of a technical nature and isn’t carried out on the human or animal body2.

The proviso “carried out on the human or animal body” is interpreted to mean that the performance of the step “implies any interaction with the human or animal body, necessitating the presence of the latter”10. No specific type or intensity of interaction is required10.

In most cases, the collection of data inherently necessitates the presence of the human or animal body, and the comparison of the data with standard values and the finding of a significant deviation are considered non-technical mathematical methods. However, there are exceptions.

Example: There is case law indicating that the exclusion may be avoided, for example, if the steps of (ii) comparing the collected data with standard values or (iii) the finding of a significant deviation necessitate complex computation using data-processing devices11.

Tips: The technical step that isn’t carried out on the body must be one of: (i) the collection of data; (ii) the comparison of the data with standard values; or (iii) the finding of a significant deviation (i.e. a symptom). Other, intermediate steps will not be considered when assessing the diagnostic character of a method2,3. Therefore, this exclusion can’t be avoided by adding additional technical steps that are not carried out on the human or animal body, if those steps don’t form part of steps (i), (ii) or (iii)12,13.

As the proviso “carried out on the human or animal body” requires interaction with the human or animal body, a method step performed by a software program within a device that is simply designed to be worn on the human body, but does not require any interaction with it, should not be interpreted as “carried out on the human or animal body”.

Summary

In conclusion, the manner in which the EPO interpret the “method of diagnosis” exclusion under Art.53(c) EPC means that it is frequently possible to obtain claims to the use of healthcare wearables which function in the diagnostic space, provided the description and claims are drafted with the exclusion in mind.

Patents and Trade Secrets: IP Protection of AI in Digital Health and Wearable Devices
January 12, 2024
Artificial Intelligence (AI) has expanded rapidly in many industries, particularly health care, ranging from radiological imaging, clinical diagnosis, drug discovery, and cardiac software that uses AI to guide users, to fitness training and equipment. Consistent with this trend, the World Intellectual Property Organization reported that the number of AI-related patent applications filed each year increased by a factor of 6.5 between 2011 and 2017.[1] The COVID pandemic has accelerated the transition to digital health, and the use of wearable devices in telemedicine has become more prevalent. In an October 19, 2023 update, the Food and Drug Administration (FDA) projected a 30+% increase of Artificial Intelligence/Machine Learning (AI/ML)-enabled medical devices in 2023 (compared to 2022).[2]

Trade Secret vs. Patent Protection

Organizations want to find the appropriate IP strategy to protect their AI technologies. Generally, trade secret protection is better suited to the source code, training data set, and internal database structure of an AI system, while patent protection may be better suited to the user interface and the complex algorithm underlying the AI system.

Other factors, such as the nature, use, and control of the AI technology, may also determine the appropriate IP regime. For example, if technology is non-public facing, used only in-house, difficult to reverse engineer, and there is a well-controlled plan in place to keep the information secret, such technology may be kept as a trade secret. In contrast, if the technology is public facing, susceptible to reverse engineering, requires data sharing, or the technology is to be licensed out or offered as Software as a Service (SAAS), then the technology is probably better protected under the patent regime.

Notably, if the technology pertains to an industry that is subject to a public disclosure requirement, patent protection may be the better choice. For example, the FDA discusses a proposed framework that would require manufacturers to provide descriptions in premarket submissions for what aspects of their products they intend to change through learning algorithms and how the algorithms will remain safe and effective as the Software as a Medical Device (SaMD) and its associated algorithms learn and change over time.[3] Accordingly, organizations should consider filing a patent application to cover the novel aspects of their AI/ML systems before they are made public to comply with regulatory requirements.

Patent Protection Hurdles

In the event that an organization decides to file a patent application to protect its AI/ML-based inventions, it will need to overcome hurdles, particularly in the area of subject matter eligibility under the Alice test,[4] as illustrated in recent court cases.

First, the claimed invention cannot merely recite black box terminologies. Courts have invalidated AI/ML-related patents where the claims and specification lack sufficient specificity regarding the invention. In PurePredictive, the court invalidated a patent that merely recited the concept of predictive analytics performed on a generic computer without more.[5] The claimed invention used AI to “generate a predictive ensemble in an automated manner with little or no input from a user or expert.”[6] The court reasoned that the patent simply recited an abstract idea, a patent-ineligible concept, and lacked sufficient specificity to meet the second step of the Alice test because its claims and specification only described “generic ‘modules’” rather than specific system architecture.[7]

In Vehicle Intelligence, the Federal Circuit similarly invalidated a patent for lack of implementation details. The patent claimed systems and methods that screened equipment operators for impairment, selectively tested those operators, and controlled the equipment if an impairment was detected.[8] The specification provided examples of such equipment, which included a wide range of vehicles and machinery, as well as examples of impairments, which included intoxication, physical impairments, medical impairments, and emotional impairment.[9] The court found that the claims were directed to the abstract idea of testing operators for impairments because the claims were not “limited to a particular kind of impairment” and did not “explain how to perform either screening or testing for any impairment, specify how to program the [system] to perform any screening or testing, or explain” how to control the equipment if the system detected an impairment.[10] As such, the court found that the claims did not provide a redeeming inventive concept because the patent’s specification and claims did not explain how the system worked or how it achieved its supposed advantages over the prior art.[11]

Additionally, patent applicants should show how their invention makes an improvement in the technological field and take care to distinguish it from mental processes, generic methods, and other abstract ideas. In the Federal Circuit case, Braemar, a patent involved AI technology in an electrocardiographic telemetry device that would monitor a patient’s cardiac activity, transmit information of cardiac events meeting the certain criterion to a remote medical receiver, and discard events that do not meet the criterion.[12] The district court found that the claims in the ‘237 patent are directed to the patent ineligible abstract idea of “collecting, classifying, and selectively transmitting relevant data” although the technology enables “accurate, automatic review of a large volume of cardiac monitoring data that was previously reviewed manually by trained technicians.” The court also held the claims recite performing the abstract idea with conventional technology and fail to provide any specific, inventive technological improvement, as opposed to generic gathering and processing activities that can be carried out manually. On appeal, CardioNet attempted to argue that the claims are directed to “improved electrocardiographic monitoring systems” and “the claimed methods increase the relevance of data presented to physicians by automatically identifying and discarding less clinically-significant events.”[13] However, the Federal Circuit stated that the purported improvement is the abstract idea of classification and filtering of data, not an improvement in the functioning of computer capabilities. The Federal Circuit then found the claims to be invalid because they did not describe an inventive concept; instead, they merely recited mental processes and routine data manipulation, which are “abstract ideas with conventional techniques.”[14]

Likewise, the Federal Circuit recently invalidated a university’s patent application directed to a computerized statistical method for haplotype phasing, which has the potential to revolutionize personalized health care and tailor treatment according to a patient’s genetic background.[15] The court found that the invention covered “abstract mathematical calculations and statistical modeling,” and that “practical, technological improvements” must extend “beyond improving the accuracy” of a mathematical calculation to demonstrate patent eligibility.[16] The court emphasized that a “claim for a new abstract idea is still an abstract idea” and that “patent law does not protect such claims, without more, no matter how groundbreaking the advance.”[17]

Conclusion

Trade secrets and patents can protect different parts of AI systems. Together, they afford more robust protections for AI/ML-based inventions. When patenting AI-related inventions, applicants should avoid using black box terminologies and instead define the technological improvements with sufficient specificity to set the invention apart from abstract ideas and generic techniques.

Patent War In Healthcare Wearables
Published Dec 20, 2023

If Apple Watch happened to be on your holiday shopping list, the news we have might have you on the move.

Here’s a glimpse into what we’re covering today:

Patent War In Healthcare Wearables

What’s Happening: Apple is pulling its Apple watch Series 9 and Ultra 2 smartwatches from the US markets to comply with an import ban issued by the International Trade Commission. This comes following its patent dispute with medical technology company Masimo over the blood oxygen feature and SpO2 sensor on the devices.

The bigger picture: Recent times have seen blurring of lines between consumer health wearables and FDA approved medical devices.

Non-medical devices provide useful insight into behaviors that may lead to better health. They also open up new avenues of opportunity for healthtech entrepreneurs looking to dabble in the world of healthcare mobile and wearable apps. However, for medically relevant readings, approved medical devices is what patients need to fall back on.

The recall comes in wake of another patent battle of Apple and AliveCor over portable ECG tech.

What’s next: Apple Watches with blood oxygen measuring capabilities will be removed from Apple’s website from tomorrow, and from its stores on Dec. 24. The older version of the Apple Watch, which tracks heart rate and physical activity, will still be available.

The ITC import ban is currently undergoing a presidential review period, and President Joe Biden will have the opportunity to veto it by Dec. 25.

In the meanwhile, Apple is working on a range of legal and technical options, if the ban holds. Engineers at Apple are moving to make changes to algorithms in the smartwatches, adjusting how the technology determines oxygen saturation and presents the data to customers, Bloomberg reported.

The bigger implication is around whether or not Apple can use the blood oxygen sensor technology that is in question on future devices, or if they'll have to reach a settlement or come up with a new solution.

MyAIWatchTM, a patent-pending technology, is the brainchild of MCICs Acting CEO
Ben Friedman, whose 40+ years as a licensed pharmacist brought unique insights
into the technology and programming behind the new smart device.

Laneassist needs what?????????? Guess - patent-pending technology,

I am thinking tomorrow!!! 5th business day since the Atty letter filling. Unless those guys at OTC M relapsed and started back watching midget porn movies. Hopefully, they have those batch programs running to update that information tonight.

And as you and I have worked really long hours for years moderating the MCIC message board, I take pride that we have consistently done a terrible job. That's cool.

Morning.

And I am looking forward to the answer. It's time........................

Yep. And this one has been too quiet about recent actions. Very few drop shell risk and go back to current info status. Why do all that and not say a word?

Where is the news on what recently happened here? Did the dog eat the press releases? Dropping shell risk and getting current with Atty letter/filings are good things.

Why hate one company like MCIC so much? Unless it's a personal matter? But still, no one cares either way. Go MCIC! To da moon. # 2 pencil. To da moon. #2 pencil.

Let's watch this start back trading next week. Should be interesting......

We were kinda bored with the lack of intelligent conversations about MCIC going Pink Current Info next week. No one seems to have noticed the atty letter and shell risk going away. Plus, we are not good pumpers at all. I will admit that much. How would we get paid?

Would you like to make money on MCIC or worry about clowns?

Let's be realistic about it? I think we would totally suck as pumpers. Short sale the dang thing.

It's all about who was right or wrong. Did a bet against MCIC go this far after the divorce settlement? It would take pure hatred after this many years. Go MCIC and hell yea we are 3 clowns. Keep it coming.

That is crazy. Almost as crazy as MCIC doing well, why would a few hate to see MCIC go up. Why? Divorce settlement was years back.......

Boom. Nice to get that out. Go MCIC.

That or they want to see MCIC fail because of previous divorces many years back. Ouch. It's only money.

They are setting up to cry. We are long ways from those prices. But why are they so concerned about past PPS predictions that were just jokes? 8 ex-wives being paid child support each month on the .30c one? Maybe, the ex-wives need 2.00 pps for better lawyers.

Let's start by getting to a penny or maybe telling the class what our true intentions are here? But I think we know that already. It's up to a random poster that is talking about MCIC just for entertainment purposes to fall into that trap? Try again. All of our posts are strictly for entertainment purposes and do not reflect any investment advice.

My MCIC trading strategy. Short it at the record low price and cover at the record high. Take out a second mortgage on all my friends houses to buy at the record high PPS. Then sell all that when the PPS is really low. I use this trading strategy for a tax loss to off set all my profits made betting on race horses. I think it is why my 8 previous wives divorced me. I still say they were wrong about me having a gambling problem. Anyway, go MCIC! Time for OTCMarkets to show it as current info and resume quotation for brokers/dealers. It's a major thing accomplished in my opinion. And my posts are just for entertainment purposes. I never claimed to give series 7 level investment advice or say that I was a marriage counselor. :)

Morning.

It's coming back from the dead. Again.

Hey if THEY or whatever pronoun they prefer, wish to pull up old posts. Let's go back to the BK judge that the end of the world prophets predicted would murder our stock. I posted countless times for that judge to light a burn barrel in his back yard for my MCIC stock certificates. I was holding all my shares because I wanted to do this cool jump in the air 360 and slam dunk my shares in his burn barrel. I don't know if I would really take investment advice from me. But the point is we are here for a good time and that makes it fun.

I thought the "eight ex-wives thing" was kinda funny. I didn't realize that this board was for serious investment statements only. But then again, the ex-wife thing SURE got a response. It's just too much fun being me.

I see the opposition is already in place to attack this one. That must mean it's going up. Thanks.

Dude I was joking about eight ex-wives and selling for .35C. And 2.00 was a typo and I apologize for that. I thought I typed .0002. I like to buy stocks high and sell low.

You could sell 250,000 share blocks at .20C minimum. That's not bad. It will only cost a new investor 50 grand for a starter position. I was looking at opening an account to purchase some funds with Fisher Investments and they require 500 grand minimum for a starter position. But they bring 35% returns yearly.

Sure will be nice to have the trading volume back and see how this moves up. Next week should be FUN.

Like it? Did you read it or is 8 payments a month not that bad of a thing these days?

MCIC @ .35C.. I am waiting on my stocks to go up so I can pay my child support payments. I need MCIC at .35C PPS in the next few weeks to make ends meet on these payments until I figure out how to deal with this financial challenge. I tried to tell the judges that I have to make two payments a week to two of my ex-wives. And I worked it out so that I pay all eight of the ex-wives monthly by paying two different ones per week. 48 weeks of payments per year. Technically, I get 4 weeks off without making payments to them because there is 52 weeks per year. Go MCIC!!!!!

Yes. The float is unchanged. Clean publicly traded vessel is ready to launch with something to hopefully support 50X higher market cap.

Morning.

Compile me a list of OTC pink stocks that dropped shell status, dropped shell risk, and went OTC Markets current info in 2024. The rule change alone years ago kicked most to the expert markets or just delisted after a while. We need a survey running. Hold over a penny when yield sign drops? It damn well should. Ben has worked for years. It's time to find out why.

AI can on using profanity in every spoken language worldwide (for those in a MCIC short position). In English, for example, swear words contain a higher ratio of so-called plosive sounds—including P, T, and K. Profane English monosyllables are especially likely to end in a plosive rather than begin with one. In German, profanity is also heavy on plosives, as well as on short vowel sounds.