Passive Monitoring of Physiological Precursors of Stress Leveraging Smartwatch Data

Developing the capability to continuously and noninvasively monitor the mental health status of individuals is a critical focus in the mHealth domain. The use of passively-generated data gathered via smart and portable electronic devices to monitor specific indicators of mental health has shown potential to serve as an effective alternative to traditional intrusive survey-based approaches to monitoring mental health remotely. In this study, we propose a remote health monitoring framework for dynamic, flexible, and scalable assessment and detection of physiological precursors of a stress response. Our method comprises a smartwatch-based system for continuous monitoring of primary physiological signals, followed by a deep neural network architecture that performs the fusion and processing of the multi-modal sensor readings. We empirically validate our system on a cohort of university-affiliated members of the military. Our findings demonstrate the effectiveness of our passive-sensing system for tracking perceived stress, the results of which can be used to obtain a better understanding of patient behavior and improve personalized treatments1.