USE OF NEURAL NETWORKS IN MEDICAL DIAGNOSIS

In recent years artificial neural networks have been popular both as a subject for research and as application tools in various domains. In this study, use of a neural network in the prediction of diagnostic pr obabilities is proposed. When the diagnostic probabilities of insulin- dependent diabetes mellitus were predicted both by linear regression a nd by a neural network in an empirical experiment, the predictions of the neural network were more accurate than those of linear regression. These results suggest that the use of a neural network should be cons idered whenever prediction of diagnosis is required.