Feb 12, 2013

Posted in Artificial Intelligence, Assisted Decision Making, Harlan/Siri, Robot, Esq.

Dr. Watson and Robot, Esq.

Share Button


I have previously blogged about IBM’s Jeopardy-winning supercomputer, and how its technology offers great promise to revolutionize the legal services industry.

IBM is already forging ahead, and transforming player Watson into Dr. Watson–it can diagnose diseases!

Watson’s first job is to help doctors decide what kind of treatment would be most effective for certain lung cancer patients. Doctors at the Maine Center for Cancer Medicine and WestMed in New York’s Westchester County will soon be able to quiz Watson through a tablet or computer, and Watson will dig through the patient’s thousand or so pages of records and provide answers in decreasing order of confidence. Watson isn’t actually making any decisions, per se, but he is making things easier by digging through tons of data, and surfacing just the important parts. It starts sometime next month

They key is that Watson is not treating the patients by itself, but offering doctors information they could not otherwise learn about.

Watson doesn’t tell a doctor what to do, it provides several options with degrees of confidence for each, along with the supporting evidence it used to arrive at the optimal treatment. Doctors can enter on an iPad a new bit of information in plain text, such as “my patient has blood in her phlegm,” and Watson within half a minute will come back with an entirely different drug regimen that suits the individual. IBM Watson’s business chief Manoj Saxena says that 90% of nurses in the field who use Watson now follow its guidance.

This is what I have called “Assisted Decision Making,” and will be the aim of Robot, Esq.

What kind of information is Dr. Watson crunching?

And we do mean a wealth of information; as IBM explains, Watson has spent the last year digesting more than 600,000 pieces of medical evidence and two million pages of text from 42 medical journals, and it has the ability to parse some 1.5 million patient records covering decades of cancer treatment history. That all takes the form of two separate “Watson-based” products to start with, one of which IBM expects to be used by more 1,600 providers by the end of this year.

Imagine Watson crawling through the U.S. Code or the U.S. Reports.

So how accurate is Watson? 90%. And how accurate are humans. 50%

WellPoint’s chief medical officer Samuel Nussbaum said at the press event today that health care pros make accurate treatment decisions in lung cancer cases only 50% of the time (a shocker to me). Watson, since being trained in this  medical specialty, can make accurate decisions 90% of the time. Patients, of course, need 100% accuracy, but making the leap from being right half the time to being right 9 out of ten times will be a huge boon for patient care.

This is similar to accuracy rates with predictive coding and spotting protected documents during discovery (80% v. 60%). Humans are not nearly as accurate as people would wish.

Amazingly, over the past two years, IBM has been able to shrink “Watson from the size of a master bedroom to a pizza-box-sized server that can fit in any data center,” while “improv[ing] its processing speed by 240%.”

My previous coverage on Watson is here, here, here, here, here, here, here, and here.

Share Button
Print Friendly