Artificial Neural Networks for the Identification of the Differences between “Light” and “Heavy” Alcoholics, Starting from Five Nonlinear Biological Variables

Abstract

This article makes a comparison among three types of evaluation systems based on a set of data composed of “heavy” alcoholics and “light” alcoholics. The three systems are: 1) a system based on genetic algorithms called BEAGLE; b) seven different types of Artificial Neural Networks; c) a metasystem called MetaNet. The technical aim was to compare the classification capability of these three systems in terms of two classes (“heavy” alcoholics and “light” alcoholics). From the results obtained, the MetaNet system stand out. Globally, it has the best result, followed by the two Artificial Neural Networks, Squash and Logicon Projection. The results obtained prove that the advanced elaboration data systems applied in the social and health fields can be employed in prevention programs having an aim to reduce the social impact of certain pathologies correlated with different kinds of dependence.