Detecting Depressive Symptoms: Activity Tracking and Chatbot Intervention

Depressive symptoms are evident in how individuals behave in their day-to-day lives, yet these signs can easily go unnoticed and treatment can be avoided due to social stigma or lack of awareness. To enhance the accessibility and efficiency of identifying depressive symptoms, there have been proposals to utilize measurements from mobile devices and wearables. However, these proposals often consolidate their focus on a single feature or measurement. In this research paper, we present a pioneering methodology for detecting depressive symptoms through the collection and analysis of multimodal measurements reflecting daily activities, including location, sleep patterns, phone usage, and duration of phone calls. Our approach involves the development of a discreet mobile application, which gathers data from participants, including location, sleep patterns, phone usage, and duration of phone calls. Additionally, our research delves into the integration of chatbot technology to address early-stage depression assessment. In addition to acquiring data from test participants, we also employed a benchmark dataset, subjecting both datasets to analysis using several anomaly detection algorithms. Our findings revealed a negative correlation between the number of visited locations and the severity of depression while using the application, and a positive correlation with the results of the PHQ-9 questionnaires and chatbot use, indicating the potential of the chatbot as a reliable tool for preliminary assessment of depressive symptoms.