Enhanced EEG Coherence Analysis for Major Depressive Disorder: Cross-Frequency Ratios and Patient Heterogeneity

<p>Major depressive disorder affects over 280 million individuals globally, yet lacks objective diagnostic biomarkers. The neural mechanisms underlying depression remain inadequately elucidated, specifically regarding functional brain connectivity patterns. To address this gap, we applied enhanced EEG coherence methods integrating traditional magnitude-squared coherence with novel frequency ratio metrics across four bands (delta, theta, alpha, beta) in 46 patients with major depressive disorder and 75 control subjects from the PRED+CT dataset. Our approach combined connectivity measures with Support Vector Machine classification and clustering methods to investigate both group differences and individual heterogeneity. Results revealed three key findings: First, MDD patients showed reduced coherence spanning all examined frequency bands, most prominently in theta and delta, with effect sizes of d = −0.34 and d = −0.28, respectively. Second, frequency ratio calculations identified significant alterations in theta–beta balance (d = −0.32, p < 0.01) and low–high frequency coordination (d = −0.14, p < 0.01). Third, Support Vector Machine classification achieved optimal performance in the beta band (83.3% accuracy), while clustering revealed two distinct connectivity phenotypes among patients. These findings suggest that depression involves both frequency-specific connectivity reductions and systematic alterations in inter-frequency balance. The identification of connectivity-based patient subtypes and high classification accuracy indicates potential for personalized treatment approaches and clinical diagnostic applications, though validation in larger samples remains essential.</p>