Abstract Background Electrocardiogram (ECG) and photoplethysmogram (PPG) signals offer moderate ability to distinguish Obstructive Sleep Apnea (OSA) patients, including those with comorbid Major Depressive Disorder (MDD), from healthy individuals. This study investigated whether analyzing these signals during specific sleep stages could improve the detection of MDD within an OSA cohort. Methods Polysomnography data from 53 participants (OSA-only, OSA with MDD, and healthy controls) was analyzed. Recordings were segmented into 5-min intervals of pure sleep stages. Results Deep-Sleep was identified as the most discriminative stage, yielding the highest number of significant features and with strong effect sizes for both ECG and PPG. Using Deep-Sleep PPG features related to vascular stiffness, distinguishing Controls from OSA was performed with 100% accuracy. For the more complex task of identifying MDD within the OSA group, combining PPG timing delays in the peripheral pulse wave and one ECG entropy feature achieved 89.33% accuracy, with an AUC of 0.91. Conclusion This single-site research demonstrates that deep-sleep-stage-specific analysis improves the power of ECG and PPG signals to differentiate between Controls, OSA, and OSA with MDD. This framework has the potential for developing an accurate ECG and PPG-based system to screen for depression in patients with sleep disorders, overcoming delays in the assessment due to long waiting periods in sleep clinics. The translation of this research requires a large, multi-site database for the evaluation of wearable devices for this application.
Shaw et al. (Thu,) studied this question.