- IBM’s Watson supercomputer is already a Jeopardy! champion, and has now embarked on a second career in medicine, working with students at the Cleveland Clinic Lerner College of Medicine at Case Western Reserve University to improve its knowledge of medical concepts in a problem-based learning environment. Eventually, Watson is destined to use natural language processing (NLP) and machine learning to help teach students and provide clinical decision support to physicians, providing cutting-edge knowledge and enormous computing power to aid their diagnoses.
Named after IBM Corporation’s first president and not the famed Sherlock Holmes sidekick, Watson is an artificial intelligence computer system capable of answering questions posed in everyday language. Combining NLP, information retrieval, knowledge representation, and machine learning to process information in an organic fashion, Watson represents a “paradigm change in how we practice medicine,” says Dr. James Stoller, chair of the Cleveland Clinic Education Institute.
Ideally, Watson will go beyond its Jeopardy! performance of giving answers to direct questions, and will be able to formulate “inference paths” towards a probable diagnosis. By feeding the supercomputer thousands of gigabytes of clinical data, including medical dictionaries, studies, electronic health records, and other literature, Watson will be able to create a portrait of the medical landscape, drawing connections between concepts, highlighting hitherto unseen patterns, and offering physicians a list of diagnostic possibilities, weighted by probability and relevance in order for doctor’s to check their own thinking against its vast database of knowledge.
“What we learned in medical school loses accuracy over time, and is doing so at an accelerating rate,” Dr. Stoller observes. But Watson can easily cope with the wealth of new information being published every year, offering physicians a powerful tool to synthesize new knowledge with their own experience and the patient’s individual needs. “It could be a powerful, easy-to-use bedside decision support tool with deep understanding that helps make sure that our lists are complete and that there’s concordance between the physician’s thinking and the decision support thinking.”
But Watson has a great deal of work to do before it is invited to attend a clinical visit. Processing the sheer volume of information being fed into the system will take time, as will cultivating its ability to understand the data, drawing connections between terms and learning the nuances that physicians only gather from experience working with patients in the field. As intelligent as the supercomputer’s guesses may be, the human body often defies standardized explanations, and a physician’s intuition cannot be replaced by a clinical computer.
Case Western students and staff will answer Watson’s questions and correct its mistakes to enable it to learn faster, hoping that someday its artificial intelligence will help save lives by returning the favor.