Rapid Screening of Alcoholism: An EEG Based Optimal Channel Selection Approach

Alcoholism is a socio-economical syndrome in which human being may lose his/her health and wealth. The paper reports a novel approach for the rapid detection of alcoholism using Electroencephalogram (EEG) sensor. The proposed method employs absolute gamma band power used as a feature and ensemble subspace K-NN used as a classifier to categorize alcoholics and normal subject. Furthermore, an Improved Binary Gravitational Search Algorithm (IBGSA) is reported as an optimization tool to select the optimum EEG channels for the rapid screening of alcoholism. The results obtained by the proposed method are compared with the optimization algorithms like a genetic algorithm (GA), binary particle swarm optimization (BPSO), and binary gravitational search algorithm (BGSA). Fitness function for these optimization algorithms is evaluated using accuracy obtained from ensemble subspace K-NN classifier. The proposed IBGSA methodology provides a detection accuracy of 92.50% with only 13 EEG channels. Thus, it is the best candidate to bridge the trade-off of detection accuracy and the number of channels used for detection.