PPD Recognition based on Portable EEG Acquisition and Unsupervised Clustering Algorithm

Depression, affecting over 280 million individuals worldwide, ranks as the second-leading global health burden. Postpartum depression (PPD), a prevalent subtype, poses significant challenges in diagnosis, treatment, and prevention. This study employs wearable three-channel electroencephalogram (EEG) devices to collect data from 52 postpartum women, extracting power spectral density (PSD) features and applying unsupervised clustering algorithms for PPD identification. Unlike traditional methods reliant on subjective questionnaire-based assessments, which are often biased by cultural beliefs or psychological barriers, the proposed EEG-based approach offers a quantitative and objective evaluation. The lightweight, portable three-channel EEG system enhances practicality, enabling versatile applications in diverse clinical and community settings. Results demonstrated clear clustering of high- and low-risk participants, with silhouette coefficients reaching 0.778 and Davies-Bouldin indices as low as 0.221, suggesting high internal consistency and separability among EEG-derived features.