A P300 potential evaluation wavelet method comparing individuals with high and low risk for alcoholism

Electrophysiological recordings of event-related potential P300 reveal transient information contained in electroencephalogram, helping to diagnose individuals that are predisposed to alcoholism. Generally, this component has an amplitude significantly smaller in patients at high risk of developing the disease than in low risk ones, being a major endophenotype of the disease. In this work, we propose an alternative system to automatically classify P300 signals of individuals with high risk and low risk, composed by two modules: a discrete wavelet transform that extracts features and an artificial neural network module which identifies the patterns. After training, 97.36 % of correct classification was obtained in the database from Collaborative Study on the Genetics of Alcoholism.