Theranostics:
My problem with it is more fundamental.
The first step in "theranostics" is to run some genomic / proteomic profile to see what type of therapy will be prescribed.
Consider that the people working to devise the genomic / proteomic "fingerprints" for certain diseases are trying to develop them using samples from patients with the disease. Well, if you were able to reliably determine the disease in a patient without these expensive tests (and hence enroll them into your study), then why do we need the tests? Why add 100s of dollars of unnecessary spending on top of a currently valid and reliable diagnosis?
Second of all, the most obvious markers are clear and provide symptoms. The additional genetic / proteomic testing is infatuated with sensitivity and the ability to detect more people with disease based on little known, but new markers. Colour me unimpressed if the aim is to actually ignore the biggest indicators of disease in an effort to validate the tertiary and quaternary bit actors.
Until we get better at treating disease, theranostics is only going to result in the accumulation of petabytes of genomic and proteomic profiling data (many of it acquired to different standards and different levels of accuracy and reliability) that will simply become a means to retroactive validate a breakthrough on the treatment side.
Market Data 
Markets 
AI
