Medical Diagnostic Decision Support

Medical diagnostic decision-making is a complex task that consists in finding the right diagnosis from the signs and symptoms presented by a patient. Computers have rapidly been considered as potential diagnostic aids in medical decision-making. This chapter first presents medical diagnostic modeling as a hypothetico-deductive reasoning process. Then, the different approaches developed to provide computerized medical diagnostic decision support are proposed. Initial numerical approaches, either statistical or probabilistic, are first presented. Examples of clinical scores, more recently developed, are given. Then, medical expert systems are described. The three components of an expert system, the knowledge base, the base of facts, and the inference engine, are introduced. A focus is given on knowledge representation formalisms with the description of production rules, decision trees, semantic networks, and frames. The subsection describing the inference engine starts with a presentation of the three types of inference (deduction, induction, abduction). The principles of formal logic are given and the main ways the inference engine may operate are described (forward and backward chainings). Finally, historical medical expert systems, such as Mycin and Internist, as well as systems currently available for medical diagnostic decision support (DXplain™) are described.