Eeg-Based Prediction of Major Depressive Disorder With Hybrid Machine Learning Models and Synthetic Data Augmentation

Major Depressive Disorder (MDD) is a prevalent yet challenging condition to diagnose due to its reliance on subjective clinical assessments. This study presents a novel hybrid machine learning approach with electroencephalogram (EEG) data to enhance diagnostic accuracy and interpretability. The MODMA dataset is used alongside synthetic data generated through Conditional GANs and Gaussian Mixture Models to improve generability. A comprehensive feature extraction process captures linear, non-linear, and time-frequency characteristics of EEG signals. The proposed hybrid model integrates multiple machine learning algorithms through majority voting, achieving robust and accurate MDD classification.