Differentiating depression subtypes through three-electrode EEG and emotional stimuli: a machine learning approach

Clinical Snapshot

Key Findings

• P Value: Not reported for diagnostic accuracy

• Effect Size: 83.1% accuracy for HC versus FDE discrimination

• Primary Outcome: Diagnostic accuracy for depression subtype classification

• Nnt Or Sensitivity: 70.6% accuracy for three-class classification (HC vs FDE vs RDE)

• Confidence Interval: Not reported

Clinical Application

Limited feasibility requiring specialized VR equipment, EEG setup, and machine learning infrastructure not readily available in routine clinical practice Would require TGA approval for diagnostic use. Not currently supported by Medicare or PBS. Limited applicability in Australian primary care or standard psychiatric practice due to resource requirements. Adults with depression requiring subtype classification, though selection criteria unclear

Abstract

We propose an EEG-based framework for depression subtype assessment using emotion-modulated neural dynamics elicited by immersive virtual reality (VR). EEG was recorded from 70 participants (31 first depressive episode, FDE; 18 recurrent depressive episode, RDE; 21 control participants, HC) using a compact frontal montage (Fp1/Fpz/Fp2) during positive and negative VR conditions, focusing on the immediate post-stimulus regulation period. Subject-level features were analyzed with a linear mixed-effects model to probe group-by-condition interaction patterns, and a Bi-Emotional Siamese Network (BESN) was developed to model baseline-anchored within-subject deviations by fusing positive/negative reactivity streams with path-signature-based temporal encoding. Feature-level analyses revealed interaction signals suggestive of differential emotion-related modulation across groups. In subject-wise evaluation, BESN achieved 83.1% accuracy for HC versus FDE discrimination and 70.6% accuracy for three-class classification (HC versus FDE versus RDE), outperforming conventional machine-learning baselines. Robustness was further supported on an external public resting-state EEG dataset (MODMA), achieving 78.2% accuracy. These results suggest that baseline-anchored, emotion-modulated EEG dynamics combined with path-signature modeling provide an objective and generalizable computational approach for depression-related group discrimination.

References

1. 1.Liang, G., Zhang, X., Yang, Z., Wang, Y., Liao, D., Guo, J., Zhang, L., & Xu, X. (2026). Differentiating depression subtypes through three-electrode EEG and emotional stimuli: a machine learning approach. Biomedical Physics & Engineering Express. https://doi.org/10.1088/2057-1976/ae55a7