A model of differential diagnosis in histopathology using the theory of hypergraphs and social choice

Abstract

A model of differential diagnosis based upon the notion of a minimum vertex cover of a hypergraph and the Kemeny social choice function is described. It presents a coherent strategy for dealing with qualitative descriptions of incomplete and uncertain knowledge, such as are commonly found in the medical domain. A description of how the hypergraph model may be embedded within a decision support system is given, and the performance of this system when applied to the problem of differential diagnosis in breast histopathology is discussed. A comparison is made between the hypergraph model and set covering model (as proposed by Reggia), illustrating the ‘dual’ nature of their relationship. The problem of ranking a set of diagnostic alternatives is explored using techniques taken from the theory of social choice. Finally, the possible areas in which the hypergraph model may be extended are discussed.