Use of artificial neural networks for clinical diagnosis

The application of two unsupervised Artificial Neural Networks (ART2 and Kohonen) for the classification of patients in three medical fields is considered. In the first case, data obtained from spectral analysis of the EEG signals are used in order to detect brain disfunction of persons exposed to organic solvents. The Contingent Negative variation which is an Evoked Response is utilized in the last two cases for the diagnosis of Huntingdon's Disease and for the separation of schizophrenic subtypes.