Automating Judgmental Decision Making for a Serious Medical Problem

Abstract

A decision theory model of the diagnosis and treatment of a serious medical condition, acute renal failure, has been implemented as part of a study of automated judgmental decision-making. An interactive diagnostic computer program was used to test the model. This program employs formal decision procedures and the subjective assessment of likelihoods and preferences of experts to analyze the diagnostic/treatment problem. The program's success in duplicating the decisions of expert clinicians in about 90 per cent of the cases used indicates that the method of analysis is not only a convenient structure for theoretically describing diagnosis and treatment, but that it is potentially a practical way of analyzing such decision problems.