Classifying subclinical depression using EEG spectral and connectivity measures.

Area of Science:

• Neuroscience
• Psychiatry
• Biomedical Engineering

Background:

• Early detection of depression is vital for preventing severe episodes.
• Subclinical depression, or dysphoria, often precedes major depressive disorder.
• Electroencephalography (EEG) offers a potential biomarker for neurological and psychiatric conditions.

Purpose of the Study:

• To develop an automated classification pipeline for detecting subclinical depression (dysphoria) using EEG signals.
• To identify key spectral and functional connectivity features indicative of early-stage depression.
• To evaluate the diagnostic accuracy of the proposed EEG-based detection method.

Main Methods:

• Recorded resting-state EEG signals from female participants with dysphoria and healthy controls.
• Extracted spectral and functional connectivity features from EEG data.
• Employed a nonlinear Support Vector Machine (SVM) classifier with Recursive Feature Elimination (RFE) for classification.

Main Results:

• Achieved a maximum classification accuracy of 83.91% using a combination of spectral and connectivity measures.
• Identified that 4 key functional connections in the theta band were highly informative, yielding 76.11% accuracy.
• Highlighted the significant role of cortical connectivity in the theta band for early depression recognition.

Conclusions:

• The proposed EEG-based pipeline effectively detects subclinical depression (dysphoria).
• Cortical connectivity in the theta band is a promising indicator for early depression diagnosis.
• This approach can aid clinicians in identifying and managing early-stage depression.