Uninsured patients who depend on safety-net institutions for health care have long lacked access to specialists. Now the leaders of an ambitious initiative that uses artificial intelligence to help doctors improve diagnoses will work with some of the nation’s top medical groups to close coverage gaps for low-income Americans.
Earlier this month the Human Diagnosis Project (“Human Dx” for short), an effort to build electronic consult services by “combining collective intelligence with machine learning,” announced a long-term partnership with the American Medical Association, American Board of Medical Specialties, and the Association of American Medical Colleges, among others.
The experienced physician doesn’t “think” like a Watson supercomputer. For example there are about 150 textbook causes of abdominal pain. The good doctor does not systematically go through the list of 150 to make a diagnosis. In fact doctors use inductive rather than deductive reasoning. The recognize patterns of symptoms and quickly narrow the differential diagnosis. Except for the rare case AI not all that useful.