Mental stress is a significant health concern in the modern society and plays a crucial role in influencing brain function as well as several chronic health conditions. Stress and mental workload lead to various physiological changes, affecting heart rate, skin temperature, and sweat gland activity. The accurate detection of mild stress is important in order to better understand these effects.
In this project, we conduct a study where we use wearable devices to extract multimodal physiological signals from users, thereby quantifying cognitive load and distinguishing mild levels of stress. We induce mild stress and different levels of cognitive load in the participants in a "controlled" laboratory setting as well as in a real-life, "uncontrolled" environment. We use the Muse 2 headgear to collect EEG data and the Empatica E4 wristband to collect Heart Rate Variability, Skin Temperature, and Galvanic Skin Response. We also extract accelerometer data from both devices. These signals have been chosen for our study because of their non-invasive data acquisition, low-cost setup, ease of use and high temporal resolution.
We aim to determine whether either of these modalities alone can distinguish different levels of stress and mental workload for a relevant yet challenging task - reading and summarizing of texts. We extract suitable features from the respective signals during the experiment and feed them into chosen classification models to distinguish mild levels of stress. There is very little evidence of the reliability of these sensors in ambulatory or real-time settings and therefore, we also aim to bridge this gap in the existing literature with our research.