Computer Based Detection of Alcoholism using EEG Signals

Abstract

Electroencephalogram (EEG) signals specify the electrical signals caused by neurons, which are very helpful to discriminate different activities of the human brain. In order to distinguish between the various functions of the human brain, electroencephalogram (EEG) signals describe the electrical signals produced by neurons. Traditional time domain or frequency domain methods of analysis are useless for any application since these signals exhibit nonstationary features. In this study, we examined a variety of more sophisticated time-frequency-based EEG signal component extraction algorithms for grouping and using it for automatic alcohol detection. The machine learning algorithms for alcohol EEG detection are proposed in this paper. It is made sense to conduct a thorough analysis of the decay of signs into recurrence subgroups using wavelet approach, DWT, and a set of quantifiable highlights that were subtracted from the EEG signals to address the circulation of wave coefficients. Furthermore, techniques like ICA and PCA are utilized for decreasing the feature vector dimensions, also an aspect of information and sign vectors which can be changed over completely to highlights vectors. Finally, a linear discriminant analysis (LDA) based classifier has been used after information decrease by reasonable determination technique and the classification performance has been measured by the parameters such as specificity, sensitivity and accuracy. These values in our work are 98.9%, 98.2% and 98.7% respectively.