Detection of Schizophrenia at the Onset from EEG Signal - A Machine Learning Based Approach

Abstract

The first signs of schizophrenia are thought to manifest during late adolescence. Hence, if the diagnosis can be made during the onset, then the patient can lead a comparatively functional life. The most cost-effective way to monitor the brain activity is using electroencephalography (EEG). Since the visual analysis of EEG comes with interpretation issues, researches are being carried out for machine learning based interpretation system. The authors proposed classification models using several machine learning algorithms to distinguish between normal and schizophrenic subjects from EEG data taken during the resting phase. The best result was by Random Forest (RF) with precision, sensitivity, and specificity of 0.965, 0.965, and 0.95 respectively.