The full text of this item is not available at this time because the student has placed this item under an embargo for a period of time. The Libraries are not authorized to provide a copy of this work during the embargo period, even for Texas A&M users with NetID.
Detecting and Managing Stress Using Off-the-Shelf Computing Devices
Abstract
Stress is an unavoidable part of the human experience. While small doses of stress can help one focus and improve productivity, being constantly exposed to stressors can lead to deleterious health outcomes. As such, identifying stress episodes early on and managing stress with relaxation interventions can help alleviate some of its negative consequences. Traditional stress detection approaches (e.g., measurements of stress hormones, psychometric instruments) can be difficult to implement in modern workplace settings due to logistic constraints, and they only provide a single-point measurement.
To address this issue, the first study in this dissertation proposes automatic and unobtrusive stress detection methods that leverage devices that are already part of the modern workplace setting and can be used as is, or easily instrumented with little cost. Specifically, the approach consists in augmenting keyboards and mice with sensors to measure typing pressure and mouse gripping pressure, respectively. Using a method based on linear discriminant analysis and k-nearest neighbors, I found that when pressure information is added to keystroke dynamics and mouse trajectories, it improves model performance by 6 and 3%, respectively.
One of the challenges in conducting research in the field of stress analysis is the lack of publicly available datasets. To address this issue, this dissertation also presents a stress analysis dataset to be released to the public that contains typing information (i.e., keystroke dynamics) and physiological data collected while participants complete knowledge tasks under neutral or stress mental conditions. Preliminary analyses on this dataset indicated that (1) the experimental protocol elicited the desired stress response, and (2) some keystroke dynamics were significantly affected by changes in mental state.
Finally, this dissertation also investigates the use of casual games and biofeedback to deliver an engaging relaxation micro-intervention, aiming at reducing the high attrition and dropout rates observed in traditional relaxation techniques, such as deep breathing. In the approach, the game adapts in real time to participant’s breathing rate using the concept of negative reinforcement instrumental conditioning. In a 3-day ambulatory study, the game biofeedback approach outperformed a conventional intervention (paced breathing) in terms of adherence to treatment, perceived relaxation, and transfer of breathing skills.
Subject
stress detectionstress management
machine learning
keystroke dynamics
mouse dynamics
game biofeedback
skill transfer
Citation
Da Cunha Silva, Dennis (2023). Detecting and Managing Stress Using Off-the-Shelf Computing Devices. Doctoral dissertation, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /200141.