Recognition of Major Depressive Disorder via Alpha Power and Theta Asymmetry in Publicly Available EEG Datasets Employing Machine Learning Approaches

A person’s thoughts, actions, and emotions may be profoundly impacted by despair, also known MDD (as Major Depressive Disorder). Presently, millions of individuals are impacted by MDD, making it a prominent condition. Depression is diagnosed using a questionnaire rather than any objective criterion. This research presents a method for diagnosing depression using features taken from EEG signals. Features employed included alpha, alpha1, beta, delta, and theta supremacy, as well as theta asymmetry. Theta asymmetry, alpha1, and alpha2 were also used as features. Feature selection using feature combination was done using MCFS. SVM, LR, NB, and DT were the classifiers that were used. In every classifier that was tested, alpha2 outperformed alpha1 and alpha power in terms of classification accuracy. The t-test revealed a statistically significant difference between the left and right hemisphere theta powers of healthy individuals, but no such difference in depressed individuals. The average theta asymmetry is greater in healthy individuals compared to those with MDD, however this difference is not statistically significant. The best classification accuracy in SVM was 89.46% when alpha2 and theta asymmetry were combined.