A medical reasoning program that improves with experience

A physician's problem-solving performance improves with experience. The performance of most medical expert systems does not. The author has developed a diagnosis program for coronary disease that improves its performance by remembering and learning from cases that it has already solved. The program diagnoses commonly-seen problems efficiently by recalling similar, previous cases and adapting their solutions through simple modifications. When it lacks experience in solving a particular type of problem, the program resorts to reasoning from a physiological model, then remembers the solution for future use. The program can produce solutions identical to those derived by a model-based expert system for the same domain, but with an increase of two orders of magnitude in efficiency. The method described is independent of the particular domain and should be generally applicable.