Development and validation of a computer program using Bayes's theorem to support diagnosis of rheumatic disorders

Abstract

A computer program has been developed to support the differential diagnosis of major rheumatic disease categories. Knowledge for the program was derived from the sensitivity and specificity of findings in a group of 1052 consecutive patients of a rheumatological outpatient clinic. Computer predictions based on two variations of Bayes's theorem were compared with 'gold standard' diagnoses in 570 test cases from the same clinic. The results demonstrate the feasibility of this approach, but also that refinements of both the method of calculation and the knowledge are required.