APPLICATION OF THE FUZZY ARTMAP NEURAL-NETWORK MODEL TO MEDICAL PATTERN-CLASSIFICATION TASKS

This paper presents research into the application of the fuzzy ARTMAP neural network model to medical pattern classification tasks. A number of domains, both diagnostic and prognostic, are considered. Each such domain highlights a particularly useful aspect of the model. The firs t, coronary care patient prognosis, demonstrates the ARTMAP voting str ategy involving 'pooled' decision-making using a number of networks, e ach of which has learned a slightly different mapping of input feature s to pattern classes. The second domain, breast cancer diagnosis, demo nstrates the model's symbolic rule extraction capabilities which suppo rt the validation and explanation of a network's predictions. The fina l domain, diagnosis of acute myocardial infarction, demonstrates a nov el category pruning technique allowing the performance of a trained ne twork to be altered so as to favour predictions of one class over anot her (e.g. trading sensitivity for specificity or vice versa). It also introduces a 'cascaded' variant of the voting strategy intended to all ow identification of a subset of cases which the network has a very hi gh certainty of classifying correctly.