Detection of Major Depressive Disorder Using Genetic Algorithm for Features from EEG Signals

Abstract

Introduction: Depression is a widespread mental disorder that significantly impacts individuals' well-being and quality of life. In the previous few years, researchers have looked into the potential of electroencephalogram signals in detecting and diagnosing depression. The present study investigates an approach for depression detection using a genetic algorithm to optimize the selection of discrete wavelet transform features from EEG signals. The proposed method involves decomposing EEG data into seven sub-bands using the DWT, extracting relevant statistical, geometric, and physiological features, and then employing a genetic algorithm to identify the most informative features for depression recognition. The SVM classifier achieved the highest overall accuracy of 91.73%, a sensitivity of 91.01%, a recall of 92.18%, and an F1-score of 91.45%, outperforming the other models. Experimental results indicate that the proposed approach outperforms hand-engineered methods and highlighting its potential as a complementary tool for depression diagnosis.