Replies to post #115066 on Biotech Values

[DewDiligence]

The Watson program could excel in medical billing and coding. It's a growth industry, although it may be dull, tedious, and frustrating, requiring great organizational skills and patience as the go-between amongst the patient, doctor, and insurance companies.

This is a focus of 3M’s healthcare business—see http://finance.yahoo.com/news/3M-and-Nuance-Team-to-Deliver-bw-1976233925.html?x=0&.v=1 .

[DewDiligence]

IBM’s ‘Watson’ Beats Top Students at Jeopardy—Barely

http://www.bostonglobe.com/business/2011/10/31/ibm-watson-edges-out-students-jeopardy-barely/Rspd4LgG7j60UZjXBP643J/story.html

›By Hiawatha Bray
November 01, 2011

IBM Corp.’s Watson supercomputer gained worldwide fame when it beat champion players of the TV game show “Jeopardy.’’ But at Harvard Business School yesterday, the legendary supercomputer almost met its match.

Almost.

Watson triumphed in a hard-fought contest against six students; three from Harvard Business School and a trio from the Massachusetts Institute of Technology’s Sloan School of Management. The human-versus-machine matchup proved to be unusually close, clinched by Watson on the final question.

“I had no idea what was going to happen,’’ said IBM fellow David Ferrucci, one of Watson’s chief developers.

The scientists and academics who arranged the showdown had more than entertainment in mind. They hoped the demonstration would provide students with a better understanding of Watson’s power and its potential to help humans solve the most complex problems. “This is about empowering the decision maker,’’ Ferrucci said.

Watson is actually a stack of 90 powerful IBM server computers housed at the company’s Yorktown Heights, N.Y., lab, represented at yesterday’s event by a video screen and a digital voice. Its vast memory contains the equivalent of a million books. More important, its software is capable of recognizing the subtle concepts buried in everyday human speech and writing.

The result, said Ferrucci, is a computer that can comprehend documents that would once have been meaningless to a machine. Armed with this knowledge, Watson can provide vital decision support for workers who must make complex, difficult choices, like a doctor prescribing treatment or a financial planner working up an investment strategy.

“Where you see decision-making based on large volumes of natural language content,’’ said Ferrucci, “this is an opportunity for this class of technology.’’

In a Jeopardy game, Watson gives only one answer - usually the right one. In yesterday’s match, the computer never gave an incorrect answer.

In real-world settings, a Watson-like system would give a user several possible answers, leaving the final decision to a human. But since Watson searches millions of documents, it will base its suggestions on far more data than any person can access.

“What humans do you know who can fit millions of books in their head?’’ Ferrucci said. “That’s what we want to do ultimately with this technology.’’

Nuance Communications Inc., of Burlington, made some of the software behind another voice-recognition system that made a splash this year: Siri, the spoken-command assistant used in the iPhone 4S, the latest generation of Apple Inc.’s popular smartphone.

The company also makes speech-recognition systems used in health care and is working with IBM on an effort to use Watson technology in medical diagnostics. The goal is a system that could instantly examine millions of patient records to discover similar cases and help doctors make better diagnoses [#msg-60068913].

“There seems to be a lack of knowledge at the physician’s fingertips, right at the point of care,’’ said Joe Petro, senior vice president of research and development in Nuance’s health care division. “We’re trying to move Watson in a direction where it can solve this type of problem.’’

Still, Watson’s impressive performance worried Andrew McAfee, principal research scientist at MIT Sloan’s Center for Digital Business. He hailed the computer’s power but warned that it and other new digital technologies are adding to the nation’s unemployment crisis by enabling machines to do jobs once performed by people.

“These incredibly powerful new computer technologies that we’re coming up with are eliminating more jobs that they’re creating,’’ McAfee said, adding that to compete, the nation’s workforce will need to become more creative and innovative, talents which Watson still does not possess.

Still, the IBM supercomputer is very good at playing Jeopardy. The computer rang up a winning score of $53,601, compared to $42,399 for the Harvard team and a dismal $100 for MIT [LOL].

The game was remarkably close. Watson fell behind the Harvard team late in the contest, but recovered. By the final question, Watson’s lead over Harvard was a mere $1,800. Both teams got the correct answer, but Watson made a huge $25,001 bet, giving it a decisive lead.

On finishing second, Genevieve Sheehan, a second-year Harvard Business School student who once competed on the real TV show, said, “I think finishing with above a $40,000 score is nothing to sneeze at.’’

MIT Sloan student Ari Oxman freely praised the machine that had conquered his team. “Watson was pretty awesome,’’ Oxman said. “It’s like Google, but better.’’

What about that winning final question? MIT didn’t know. Harvard did. The clue was a hint about a national monument: “Finding the spot for this memorial caused its creator to say, ‘America will march along that skyline.’ ’’

Watson’s answer: “What is Mount Rushmore National Memorial?’’

Right, of course.‹