IBM continues to expand the use of its Watson supercomputer from winning Jeopardy to handling incoming call-center questions to guiding cancer doctorsat Memorial Sloan Kettering to better diagnoses. Today it announced a new pilot program for Watson at Houston’s renowned MD Anderson Cancer Center. The institution has been trying out Watson for a little under a year in its leukemia practice as an expert advisor to the doctors running clinical trials for new drugs.
According to the FDA, some $95 billion is spent on clinical trials each year and only 6% are completed on time. Even at the best cancer centers, doctors running trials are feeling around in the dark. There are so many variables in play for each patient and clinicians traditionally only look at the few dozen they feel are the most important. Watson, fed with terabytes of general knowledge, medical literature and MD Anderson’s own electronic medical records, can riffle through thousands more variables to solve the arduous task of matching patients to the right trial and managing their progress on new cancer drugs.
“It’s still in testing and not quite ready for the mainstream yet, but it has the infrastructure to potentially revolutionize oncology research,” says MD Anderson cancer doctor Courtney DiNardo. “Just having all of a patient’s data immediately on one screen is a huge time saver. It used to take hours sometimes just to organize it all.”
In a blog post published today, Dr. DiNardo gave an example of filling in for a colleague who was out of town. She had to meet with one of his patients who had a particularly complicated condition that needed a management decision. “Under normal circumstances, it may have taken me all afternoon to prepare for the meeting with enough insight to provide the most appropriate treatment decisions. With Watson, I am able to get a patient’s history, characteristics, and treatment recommendations based on my patient’s unique characteristics in seconds.”
Watson isn’t taking over the work from doctors. It’s more of an advisor, giving evidence-based treatment advice based on standards of care, while providing the scientific rationale for each choice it makes. Doctors can click on any option and drill down to the medical literature or patient data used to generate that option, along with the level of confidence Watson has in that source. “It’s tracking everything and alerts you when something’s wrong, like when you need to start a patient on prophylactic antibiotics if they have severe neutropenia (an abnormally low number of a certain type of white blood cell),” said Dr. DiNardo. “We might know that here at MD Anderson because we see hundreds of leukemia patients, but a less expert center might not. Now everyone in the world becomes a leukemia expert.”
So far only ten oncologists on faculty at MD Anderson have been involved in the pilot and, while Watson is using real patient data, its advice hasn’t been applied to real patients yet. That might come in early 2014. But for DiNardo, just the Big Data opportunity alone of using Watson to look across all of genomics and molecular research while assimilating every structured and unstructured byte of patient data, is tantalizing. “The potential is really amazing,” she says.