Schizophrenia Detection Based on Electroencephalogram Using Support Vector Machine

Abstract

Schizophrenia is a mental disorder caused by genetic factors and brain chemical factors. This disease requires early treatment. One way to detect schizophrenia is to use an electroencephalogram (EEG). An EEG is a device used to record signals generated by the brain’s electrical activity. This study was conducted on detecting Schizophrenia brain disorders based on EEG signals using the Alexnet Convolutional Neural Network (CNN) algorithm with SVM. CNN is a popular algorithm and state-of-the-art in machine learning, and SVM is still the baseline for comparing the proposed new methods. The dataset used in the study was taken from 32 normal subjects and 49 schizophrenic subjects. The data consisted of 3072 features. The test results show SVM has better performance than CNN, with a maximum accuracy of SVM 0.792 in comparison with CNN accuracy is 0.76. The fastest training time is SVM 0.5 seconds while CNN is 88 seconds, CNN training time is longer because CNN performs convolution calculations on five layers.