Design and Development of Machine Learning-Based Depression Identification Decision Support System

Depression is a prevalent mental disorder characterized by a loss of interest in positive aspects of life, fatigue, and impairment in daily activities. In severe cases, it can significantly affect an individual's quality of life. This research aims to develop a system for identifying individuals with depression based on their activity data. The proposed system, named the Depression Identification Decision Support System (DIDSS), utilizes data collected from wearable sensors to extract statistical features in both Temporal and Frequency dimensions. These features are then used to train an XGBoost classifier. The classifier achieves impressive results, with an accuracy of 99.48% and a sensitivity of 99.88%. Moreover, a user-friendly DIDSS based on the XGBoost model has been implemented, allowing interactive entry of activity data for real-time depression identification.