About 6 weeks ago, IBM and Jeopardy made history by pitting a computer against human Jeopardy champions to see if IBM had created software that could “understand” the context in natural language. The project is dubbed “DeepQA”. Did you see the shows? There were three Jeopardy episodes and several on the background of creating the computer Watson. You can see the replays on Youtube.
The day after IBM Watson won the match, we had one of the team’s engineers (Marshall Schorr) visit us at the Toronto Lab to tell us about the project and to answer questions. I stood up to ask a question, but first I made the following comment “If you’re looking for a sequel, get IBM Watson to help TV’s House with his very difficult medical cases".” I’ve seen similar comments from other people. Not that it will happen, but one of the best things about the software that was created is the possibility to help diagnose problems. After the Jeopardy show, it was announced that IBM would be working with Columbia University Medical Center and the University of Maryland Medical School on health care analytics research.
Of course the intention is NOT to REPLACE a physician, but only to assist a physician as you can read in this article “Why IBM’s artificial intelligence “Watson” could not replace a physician”. According to the video Perspectives on Watson: Healthcare, 20% of medical errors are diagnostic errors. Also, medical literature doubles every 7 years, making it physically impossible for any practitioner to read and absorb the information that they need.
I heard a story about a case that took medical professionals 6 months to diagnose. They fed the symptoms to Watson and diagnosed the issue in less than 30 minutes. Can you imagine how helpful this will be?
Another interesting thing that Watson would be able to do is to harness all the data generated by patients in hospitals around the world to determine if there are mini epidemics occurring or to see trends in medicines that might have unreported side effects. See articles “Dr. Watson, Please Report to the Health Care System” and “Treat the Patient, Not the CT Scan”.
Last year at the IBM Information on Demand Conference, we had guest speaker Dr. Atul Gawande speak to us. He did a book signing for his current book “Checklist Manifesto”. I liked the book so much that I’m now reading “Complications: A Surgeon's Notes on an Imperfect Science”. The book was published in 2003, but the chapter “The Computer and the Hernia Factor” could be read as a plea to have a supercomputer like IBM Watson helping to diagnose problems.
Personally I’ve seen times in my own medical experiences as a patient, where I would have completely welcomed my doctor consulting with Watson to help figure out what was going on.