WiStress: Contactless Stress Monitoring

How does WiStress work?

At the core of WiStress’s design is a novel machine learning pipeline that can map wireless reflections to stress levels. The pipeline extracts three key stress-correlated biometrics from wireless reflections: breathing, heart-rate variability (HRV), and motion. WiStress operates in three main stages.

• Capturing the RF Reflection: In the first stage, WiStress captures the reflection off a nearby user’s body. This component can detect the presence of a user, determine when they leave and return to its sensing field (within approximately 3m radius of the device), and eliminate the impact of other users and noise in the environment. It outputs a time-domain signal corresponding to the user’s movements.
• Self-similarity-based Perception Network: WiStress’s second component takes the time-domain signal corresponding to the user’s movements and outputs the user’s heart rate variability. This component exploits a self-similarity matrix (SSM) to zero in on a user’s heartbeats, and constructs a deep learning architecture that can robustly extract these heartbeats and eliminate extraneous movements. 
• Stress Classification Module: WiStress’s final component takes the heartbeats extracted from the perception network, selects and extracts stress-related features from the heartbeats as well as breathing cycles and movements, and uses the combined features to infer a user’s stress level. Next, we describe each component of our design in detail.

How well does the system work?

We built a prototype of WiStress and tested it on 22 different subjects across different environments in both stress-induced and free-living conditions. Our results demonstrate that WiStress has high accuracy (84%-95%) in inferring a person’s stress level in a fully-automated way, paving the way for ubiquitous sensing systems that can monitor stress and provide feedback to improve productivity, health, and well-being.