A categorization methodology for the analysis of the mortality rate in psychiatric hospitals

Analyzes the data relative to the mortality rate of patients in psychiatric hospitals in the Italian Region of Latium using a neural classification methodology. Given the superior classifying ability of this methodology, the study shows how research on a limited sample (approximately 10% of the total available data) would allow for a satisfactory generalization level. The neural approach confronts two classification algorithms: a backpropagation algorithm, and one developed by the authors, in which the learning function makes use of an adaptive rule, whose main characteristic is its strong dependence on time related to the procedures of input patterns