Predicting depression severity using effective and functional brain connectivity of the electroencephalography...

Area of Science:

• Neuroscience
• Psychiatry
• Computational Biology

Background:

• Major depressive disorder (MDD) significantly impacts individuals and communities, with current diagnostic methods lacking objective biomarkers.
• Electroencephalography (EEG) analysis, particularly functional connectivity (FC) and effective connectivity (EC), shows promise in understanding brain network alterations in MDD.

Purpose of the Study:

• To identify predictive biomarkers for MDD by analyzing subtle changes in brain dynamics using EEG.
• To elucidate the neurophysiological underpinnings of MDD by examining functional and effective connectivity.
• To correlate connectivity features with depression severity.

Main Methods:

• Resting-state EEG data from 44 MDD subjects were analyzed.
• Graph-theoretical approaches were used to extract EC features (Phase Slope Index - PSI) and FC features (Weighted Phase Lag Index - WPLI).
• Correlation analysis linked connectivity features (EC diameter, FC global efficiency in alpha band) to depression severity.
• Machine learning regression models predicted depression scores using significant connectivity features.

Main Results:

• Significant associations were found between EC diameter and FC global efficiency in the alpha frequency band with depression severity.
• Machine learning models demonstrated comparable performance in predicting depression scores using both EC and FC features.
• FC features yielded slightly better prediction accuracy (RMSE: 4.71, MAE: 3.93) compared to EC features (RMSE: 5.01, MAE: 4.29).

Conclusions:

• Alterations in functional and effective brain connectivity are demonstrably linked to depression severity.
• EEG-derived connectivity features show potential as objective biomarkers for MDD.
• These findings may enhance diagnostic accuracy and inform therapeutic strategies for major depressive disorder.