Classification of EEG signals correlated with alcohol abusers

The availability of quantitative biological markers that are correlated with qualitative psychiatric phenotypes helps us utilize automatic methods to diagnose and classify these phenotypes. EEG signals are appropriate means for extraction of these quantitative markers. According to the literature, many brain disorders and/or mental tasks can be detected by analyzing EEG signals. One such a psychiatric phenotype is alcoholism. In this paper different statistical classifiers including Bayes classifier with Gaussian kernel, Bayes classifier with KNN pdf estimator, k-nearest neighbor classifier and minimum mean distance classifier have been used in order to classify alcoholics and normal people by analyzing their EEG signals. Then by applying PCA to the feature vector and reducing the number of features to only one feature it is shown that an accuracy of 100% can be achieved for separating the two classes.