PC-KANDID: An expert system for electromyography

摘要

This paper describes PC-KANDID, a heuristic expert system for electromyography (EMG) that assists the physician with diagnosing and EMG test planning within the full range of neuromuscular disorders. It has been implemented on a PC in PC-PROLOG. The system does test planning using minimal information supplied by the physician and previous test results. From a set of test results the system derives the pathophysiological state of the tested anatomical structure using data reduction of test variables and categorical handling of uncertain evidence. It then generates a list of possible disorders that can explain some or all of the findings, and categorizes them according to whether the pathophysiological manifestations support or contradict them. The physician can repeat this iterative plan-test-diagnose cycle until he is satisfied with the EMG diagnosis. The knowledge is adaptable and maintainable via knowledge base editors.