After proving its ability to process natural language on the quiz show Jeopardy, the IBM Watson supercomputer is now trying its hand at solving some of health care's toughest challenges. Clinical decision support systems based on the Watson engine may soon be coming to a health care system near you.
More on natural language and CDS news
Health IT Exchange expert discusses Watson's future in health care
How natural language affects electronic health records
Harnessing the power of natural language for clinical decision support
The artificial intelligence system affectionately known as Watson, named after IBM's first president Thomas J. Watson, is increasingly making inroads into health care delivery. The latest and most direct application of the technology comes from an initiative of the Memorial Sloan-Kettering Cancer Center in New York. Doctors there began using Watson in March to help them diagnose cancer cases, and determine the most appropriate treatments. The system compares patient complaints, symptoms and treatment histories to existing medical literature in order to determine the most likely ailment and the most effective treatment.
In September 2011, IBM partnered with insurer WellPoint to create a decision support system that would aid in the care of patients with complex chronic conditions. The system reviews the medical literature related to a patient's particular condition, factors in the patient's treatment preferences, and gives the physician information on several possible treatments, including which one is most likely to deliver the outcome the patient wants.
The system could have applications in many areas of care delivery. Josko Silobrcic, M.D., senior medical scientist for care delivery systems at IBM, told SearchHealthIT he expects Watson-based applications to be broadly commercialized throughout 2013.
Currently, there are no commercially available applications based on Watson technology. Every instance in which the engine is used in care delivery results from collaborations between IBM and care organizations. The systems are developed specifically for the needs of that provider.
We are focusing on health care because it is really screwed up, and is really in need of significant change.
senior medical scientist for care delivery systems, IBM
But Silobrcic said this won't be the business model for much longer. As Watson continues to prove itself in real-life health care settings, IBM will look to push Watson and health care into a closer relationship in a number of ways, though the exact nature of these commercial products has yet to be determined.
While Watson-based technologies are being developed for a number of industries, Silobrcic said during his keynote address at the Pri-Med 2012 annual conference in Boston, that applying it to health care first makes sense because the amount of data created is so vast, and new knowledge is developed at a rapid rate. The industry currently struggles to incorporate all this data into patient care to improve outcomes.
"We are focusing on health care because it is really screwed up, and is really in need of significant change," Silobrcic said.
He added that it is difficult for even the best physicians to assimilate new medical knowledge and apply it to patient care. The Watson engine is intended to help doctors make better use of the data that is out there without requiring them to memorize new best practice guides or clinical study findings. Because Watson can process natural language, it can make use of information recorded in free text formats such as a physician's notes, or an editorial published in a medical journal.
"Health care is dying of thirst in an ocean of data," Silobrcic said. "We have a tremendous amount of data, but most of [that] data [is] unstructured. We can take this knowledge and disseminate it in ways that were not available before."
Still, Silobrcic acknowledged it will not be as easy for Watson to duplicate its success on Jeopardy in health care. In the quiz show, there was only one correct answer, but there are potentially many effective ways to treat a patient.