Prediction of Fluctuations in Depression and PTSD symptoms in World Trade Center Responders: A Longitudinal Sensing and Ecological Momentary Assessment Study

Ecological momentary assessment (EMA) and passive sensing data from a large sample of World Trade Center (WTC) first responders experincing depression and PTSD symptoms will be used to explore the utility of traditional machine learning and deep learning models in prediction of real-time symptom change. Heart rate, physical activity, and GPS-location data will be used to predict changes in depression and PTSD symptoms measured multiple times daily for the course of one week. Traditional machine learning and deep learning models will be compared in their prediction of symptom outcomes and features most strongly associated with symptom changes will be identified.