Watson Computer Helping Doctors Diagnose and Treat Cancer

Save as Favorite
Sign in to receive recommendations (Learn more)
  • Topic:
  • Tags:

Many of you may remember Watson, the IBM computer that was on Jeopardy in 2011. Watson beat former champs Ken Jennings and Brad Rutter (Jennings won 74 games of Jeopardy; Rutter won 20 games and was never defeated by a human).

Now, as part of a collaboration with Memorial Sloan-Kettering Cancer Center and WellPoint, a health benefits company, Watson is helping doctors diagnose and make treatment recommendations for people diagnosed with cancer, including breast cancer.

So far, Watson has taken in more than 600,000 pieces of medical evidence and about 2 million pages of breast and lung cancer research text from medical journals and clinical trials. The computer has the power to sift through 1.5 million patient records representing decades of cancer treatment history in a matter of seconds and give doctors treatment options based on what’s been successful in the past, as well as results from the most recent studies. Experts have spent more than a year teaching Watson how to process, analyze, and interpret the meaning of all this information.

Their work is paying off. When doctors first tested Watson’s accuracy at making treatment recommendations for breast and lung cancer, the computer system was about 40% accurate. By the ninth test, its accuracy had improved to almost 80%.

The experts are continuing to refine the computer system. Still, many doctors are excited about the possibilities Watson offers. It can be difficult for doctors to keep up with the latest developments in cancer research and care. The Watson system would allow doctors to input a cancer’s characteristics, as well as information about the person diagnosed (age, medical history, etc.) and Watson would suggest a treatment plan tailored to the person and the cancer, using the most up-to-date medical information available. Watson even takes patient preferences into account.

Larry Norton, M.D., deputy physician-in-chief for breast cancer programs and medical director of the Evelyn H. Lauder Breast Center at Memorial Sloan-Kettering Cancer Center, is helping to test the Watson system. He also is a member of the Breastcancer.org Professional Advisory Board.

On his blog, Dr. Norton wrote about Watson:

“In Watson we have a computer system that can read and understand language, interact with human experts, and remember everything it has ever learned. And it can use this knowledge to arrive at answers to real-life questions. But, one may ask, what about the human side of the equation? Medical decisions are not just about information: they are also about judgment. As experienced physicians we need to take everything into account in arriving at a best management plan -- not just the individual patient's biology and the biology of their disease, but social, psychological, environmental, motivational, and interpersonal factors.

“It is here that the less obvious advantages to our project become paramount. These personal elements are captured in the language of our case histories, and Watson will learn them -- while remaining ignorant of patient identity. Furthermore, Watson will not make decisions, but will interact with the on-site physician in asking the right questions to help the patient and the doctor arrive at the right decision for that particular individual. The human is never left out of the conversation. And that humanity -- never before captured in guidelines or lists of therapeutic options -- is what makes this project unique: It goes beyond mere knowledge, as important as that is, by entering the realm of human wisdom.”

It will be some time before Watson and other tools like it are widely available. Still, the possibilities are very exciting.

Stay tuned to Breastcancer.org Research News for the latest information on new tools to help doctors make personalized diagnostic and treatment recommendations for breast cancer.

Was this article helpful? Yes / No

Leer esta página en español

Bco internal cc1718 banner miniad final
Back to Top