Simplifying Depression Diagnosis: Single-Channel EEG and Deep Learning Approaches.

Area of Science:

• Neuroscience
• Computational Psychiatry
• Medical Technology

Background:

• Major depressive disorder (MDD) is a chronic illness often diagnosed late, increasing suicide risk.
• Current MDD diagnosis relies on subjective methods, missing early symptoms.
• Traditional electroencephalograms (EEG) are impractical for home use; wearable single-channel EEG offers potential but faces data processing challenges.

Purpose of the Study:

• To develop a deep learning model for binary classification of MDD using single-channel EEG data.
• To evaluate the efficacy of specific EEG channels for MDD detection.
• To enable a less intrusive and more convenient wearable solution for mental health assessment.

Main Methods:

• A deep learning model, specifically a Convolutional Neural Network (CNN), was developed for MDD classification.
• Single-channel EEG data from central, frontal, occipital, temporal, and parietal regions were analyzed.
• Leave-one-subject-out cross-validation was employed on a public dataset.

Main Results:

• The CNN model achieved 90% accuracy in classifying MDD using single-channel EEG data.
• Specific channels (Fp1, F8, Cz) demonstrated high performance in analysis.
• The study successfully generated meaningful features from single-channel EEG for real-time processing.

Conclusions:

• Single-channel EEG data, processed by deep learning, shows significant potential for reliable MDD diagnosis.
• Wearable EEG technology can offer a practical and accessible approach to early mental health assessment.
• This approach could overcome limitations of current subjective diagnostic methods for depression.