Trust me, I'm a computer: Watson takes on health care challenge
Code conquers the medical journals
Blog IBM’s game show-topping Watson system is starting its first real job early next year. Fresh from its triumphant drubbing of the human race on US quiz show Jeopardy, Watson is now being outfitted to help us slack-jawed dim-witted humans live better and longer.
WellPoint, a large US health insurer, will deploy a shiny new Watson system early next year to help physicians diagnose and treat their patients.
I recently had a chance to talk briefly with the Watson team during a live demonstration at an IBM customer event. It is impressive technology, but there are still some misconceptions about what makes it so.
Watson is not just a highly optimised search algorithm. And, although it does perform a lot of database queries, that’s the not-so-interesting part of the package.
The art of conversation
The breakthrough is Watson’s ability to understand human language, ferreting out what we mean via the context and juxtaposition of our words and coming up in a very short time with an answer that has a high probability of being correct.
At an analyst event a few years ago, I was seated next to a CIO type from a very large US hospital group. We had a fascinating and eye-opening conversation that took my mind completely off of my over-cooked steak and focused it firmly on the woeful state of health care technology.
The first thing he told me was that there is no Big Book of the Right Things to Do in Every Medical Situation.
I always figured that the medical community has been at it long enough to have arrived at the best treatments for most, if not all, diseases and conditions – but I was mostly wrong.
While there is a body of generally accepted treatments, what your particular doctor does when confronted with whatever loathsome condition you bring to him depends upon what he’s been reading in medical journals in his spare time, or what his bosses dictate, or things his golf-foursome pals suggest has worked for them in the past.
And there is a lot of medical research available today – thousands of studies of various drugs and treatment regimes. Some of it will be out of date and superseded by newer studies; some results will contradict current findings; and still others might apply only to a tiny set of corner cases.
It is impossible to be fully up to speed all the time with current research. And even if your doctor is completely up to date, there are still judgement calls that need to be made between different alternatives.
This is where Watson comes in. The technology considers a set of symptoms, combines them with the patient’s unique health history and can then comb through every single medical journal, research study and empirical outcome to present the doctor with the best course of treatment.
Watson’s advice and suggestions will usually be presented in the form of probabilities, for example: “There is an 80 per cent chance it’s this, a 40 per cent chance it’s that, but also a ten per cent chance it’s this one.” The advice is likely to be accompanied by a suggested list of tests that winnow down the possibilities.
Finger on the pulse
To me, much of Watson’s value in the health care field will be in how it can cut the time it takes to find the correct diagnosis and determine which treatment offers the highest probability of success.
This type of system can eliminate a massive number of needless, costly tests
Properly used, this type of system can eliminate a massive number of needless, costly tests. It can also prevent many of the bad outcomes that arise from misdiagnosed conditions.
As these data-led technologies come into wider use, I expect to hear a lot of wailing and gnashing of teeth about how machines are taking over, and how nothing can replace the good old country doctor.
But Watson, and other technologies along these lines (IBM won’t be the only player in this segment), are tools that allow the doctor to spend more time on those crucial judgement calls and less time flicking through the pages of medical journals. ®
Sponsored: The Nuts and Bolts of Ransomware in 2016