IBM's Jeopardy Machine Dons Lab Coat at World's Largest Cancer Center

Doctors at New York's Memorial Sloan-Kettering Cancer Center will soon be getting a second opinion on how to treat cancer patients, thanks to IBM's Watson supercomputer. Sloan Kettering's Dr. Mark Kris learned about Watson while watching Jeopardy last year, but he didn't initially see a way to use it in his clinic. "The connection came when I learned that Watson could look at case histories at an institution like ours and say what were the best choices that our doctors made," he says.
Image may contain Accessories Tie Accessory Human Person Glasses Necktie and Shelf
© MEEMS 2008

Doctors at New York's Memorial Sloan-Kettering Cancer Center -- the world's largest private cancer treatment center -- will soon be getting a second opinion on how to treat its patients, thanks to IBM's Watson supercomputer.

Most people learned about Watson a year ago when it beat out two former Jeopardy champions on the long-running TV trivia show. And that's when Sloan Kettering's Dr. Mark Kris first learned about it too. But he didn't initially see a way to use it in his clinic. That realization came later. "The connection came when I learned that Watson could look at case histories at an institution like ours and say what were the best choices that our doctors made," he says.

Now, Kris is about nine months away from rolling out a system that can troll through the hospital's massive databases of research and treatment history and come up with recommendations for doctors.

"It's an amazing opportunity for us," says Kris, the chief of Thoracic Oncology Service at the hospital. "It will allow us to have an unprecedented way of collecting an analyzing information about individual patients and present a doctor and a patient with a listing of treatment options made by looking at case histories from Sloan-Kettering, the world's medical literature, and specific information from that patient."

The doctors are asking Watson to soak up all the information it can about oncology and case histories at Sloan-Kettering. It then learns about specific patients, and comes up with a ranked list of treatment options. It offers a kind of second-opinion for the consulting doctor based on its ability to sift thorough a colossal amount of data and make new associations that doctors might not see.

The compelling feature in IBM's supercomputer is that it can learn over time, says Dr. Josko Silobrcic, a senior medical scientist with IBM Research.

"Watson doesn't really give you an answer," Kris says. "It gives you a list of answers and then it ranks based on its certainly and probabilities which ones it thinks are the most likely."

At Sloan-Kettering, Watson will be used to help with lung, prostate, and breast cancer treatments. Kris envisions an analysis tool, available over the Internet, that can help doctors everywhere. "The goal is to make it available to doctors around the world," he says. "This is not a proprietary Sloan Kettering thing."

IBM is also working on a similar cancer analysis pilot with Cedars-Sinai hospital in Los Angeles, and the company thinks that Watson could also lend a hand at Citibank too.

The hardest part of the Sloan-Kettering project is going to be plugging Watson into all the different data formats used by the hospital. "A cardiogram and a PET scan are not necessarily the same thing, and translating them into ones and zeros, there's some need to do that," Kris says.

Kris says that the hospital won't have to install a big supercomputer to make it's pilot work. Instead, it will be using Watson over the Internet. "The idea is that it would come through the cloud to a laptop or an iPad," he says.