Major Depressive Disorder (MDD) is a prevalent psychiatric illness whose diagnosis still relies largely on subjective clinical assessment and self-reported symptoms. Increasing evidence indicates that depression is associated with tangible alterations in brain network dynamics. Electroencephalography (EEG), due to its high temporal resolution and non-invasive nature, provides a viable tool for identifying objective neurophysiological biomarkers of depression. In this study, we developed an artificial intelligence-assisted EEG analysis brain graph framework to differentiate individuals with MDD from healthy controls and to find underlying neurobiological patterns. Resting-state EEG recordings from two independent public datasets were analyzed using automated feature extraction and brain connectivity modeling. The system achieved classification accuracies of 97% and 91% on the two datasets, respectively. Beyond classification, our study aimed to understand how the AI learns from brain activity patterns and what parameters drive its diagnostic decisions.

The results of these analyses show altered frontal dynamics, diminished occipito-frontal and occipito-parietal integration, impaired frontal-posterior synchronization, reduced information integration, and large-scale brain network dysconnectivity. These findings demonstrate that AI-assisted EEG analysis can uncover clinically meaningful brain signatures of depression and provide insights into its underlying neurobiological mechanisms. These insights may help clinicians better understand and diagnose the disorder.

Soujanya Hazra is currently a PhD student in the Department of Electrical Engineering at the Indian Institute of Technology, Kharagpur. His research interests are in functional brain network modeling, multimodal neuroimaging, and AI in healthcare. His doctoral thesis focuses on AI-driven frameworks to examine brain connections in neuropsychiatric disorders.