Feed

Mental disorders, such as anxiety and depression, affect approximately 900 million people worldwide, posing severe challenges to healthcare systems and society. Accurate classification of mental disorders is crucial for effective treatment. However, current diagnostic methods primarily rely on behavioral observation and self-reported questionnaires, which are highly influenced by patient subjectivity and physician expertise. Sleep provides a stable physiological state largely unaffected by subjective emotions. Sleep-related vital signs, such as respiration and heart rate, offer valuable insights into mental health conditions. Therefore, in this study, we propose a novel method for mental disorder classification by monitoring physiological signals during sleep. We utilize a millimeter-wave radar to monitor respiratory and body movement patterns, along with a pulse oximeter to acquire photoplethysmography (PPG) signals. Statistical features extracted based on medical prior knowledge are then input into a deep neural network together with raw physiological signals for mental disorder classification. Experimental results on a real-world dataset of 447 participants validate the effectiveness of our proposed method. This study provides a portable and objective solution for mental disorder classification, contributing to improved diagnostic accuracy and facilitating broader access to mental healthcare resources.Clinical Relevance- This study provides an objective and portable method for classifying mental disorders, which is of significant importance for improving diagnostic accuracy and promoting the decentralization of healthcare resources.