Since IBM's artificially intelligent computing system, took on "Jeopardy!" in 2011, Watson has been studying hard. Enrolling in med school, Watson—like many midlife career changers—has emerged with a radically stepped-up game. Powering the New York Genome Center, Watson is leading researchers to new drug therapies, and matching patients to clinical trials. Rob Merkel, healthcare and life sciences leader for the IBM Watson Group, talks about how Watson may make your future healthier.
First, what exactly, is Watson?
Merkel: We call it cognitive computing, and the best analogy is that it's like a human brain. What we are trying to achieve with Watson are profoundly deeper insights that can have an impact on the whole healthcare ecosystem, whether that's the evolution of the way care is delivered or the explosion of information in the sciences.
So how does it work?
Merkel: Clinicians and researchers can't stay on top of all the information that's out there. Depending on the studies you look at, medical literature doubles anywhere from every two to five years. There are 700,000 new scientific articles every year. There is no way man alone could ever keep up. But Watson can. It combines all that information, and can then answer questions.
With Jeopardy!, you could ask a very tough question and get a very good answer. It's been three years since then, and we've moved beyond short answers, and can now have a dialogue with Watson. Think of it as collective intelligence, doing things that were previously impossible.
Can you give us some examples?
Merkel: We taught Watson medical policy, and now we are working with WellPoint, the insurance company, on preauthorization. So when Watson gets a request for approval on a procedure, it can read the entire patient case, compare it to policies, and see if it is in compliance. If it reaches that opinion with "great confidence," it can approve the procedure. If its confidence level is only "high," it can route it to a person, and ask if he or she agrees.
We are working with the Mayo Clinic on matching patients to clinical trials. Watson reads the patient file at the point of care, and then can quickly track down information on which trials they may potentially qualify for, information they need very quickly. With the Baylor College of Medicine, Watson has been useful in the study of p53, a protein that inhibits tumor growth. Before Watson, it would take researchers all year to come up with a single new avenue of research for this particular protein. Watson found seven within several months.
Can Watson think like a doctor?
Merkel: Yes, we're working on clinical ideas, including a leukemia project with MD Anderson, and something called Watson Paths with the Cleveland Clinic, which helps come up with differential diagnoses. And with Watson for Oncology Solutions, with Memorial Sloan Kettering, Watson reads files of patients with lung, breast, colon and rectal cancers, then recommends treatment plans. To us, this is really exciting, because there is a talent shortage. So it's an opportunity to say, "If you were going to create a model physician, what would that look like?" We're training Watson, and working with the best doctors—who treat 30,000 cancer patients a year—to do so. And now patients around the world have access to this broader talent that before now, quite honestly, was limited to the Upper East Side of New York City.