Medical Diagnosis Using Bayes Theorem

Abstract

A computerized study of the applicability of Bayes theorem to the differential diagnosis of liver disease has been made. Statistical independence of symptoms is not presumed. The semantic obstacle involved in precise definition of the symptom and disease categories is discussed. Input for the study was obtained from patient records, and diagnosis supported by tissue examination, either at autopsy or by biopsy. Correct diagnosis rate is considered sufficiently high to warrant further investigation.