Date of Award
2020
Document Type
Thesis
Degree Name
Master of Science (MS)
Department
Electrical Engineering
First Advisor
Ryan S. McGinnis
Second Advisor
Jeff Frolik
Abstract
The ability to monitor physiological parameters in an individual is paramount for the evaluation of physical health and the detection of many ailments. Wearable technologies are being introduced on a widening scale to address the absence of low-cost and non-invasive health monitoring as compared to medical grade equipment and technologies. By leveraging wearable technologies to supplement or replace traditional gold-standard measurement techniques, the research community can develop a deeper multifaceted understanding of the relationship between specific physiological parameters and particular health conditions. One particular research area in which wearable technologies are beginning to see application is the quantification of physical and mental stress levels in individuals through brainwave and physiological feature monitoring. At present, these methods are time consuming, invasive, expensive, or some combination of the three.
This thesis chronicles the development and application of a novel open source wearable sensing platform to the field of stress and fatigue estimation and quantization. More specifically, the garment in its current configuration monitors heart rate, blood oxygen saturation, skin temperature, respiration rate, and skin conductivity parameters to explore the relationship between these parameters and various self-reported stress measures. Utilizing machine-learning methods, subject-specific models were generated in an n=1 study which predicts the self-perceived stress level of the subject with an accuracy of between 92% and 100%. The garment was developed with a modular interface and open source code base to allow and encourage reconfiguration and customization of the sensor array for other research applications. The dataset generated in this effort spans the early stages of the COVID-19 pandemic as the subject experienced increasing levels of isolation and tracks physiological parameters across two months via daily measurements.
Language
en
Number of Pages
123 p.
Recommended Citation
Hewgill, Blake, "Open Source Quantitative Stress Prediction Leveraging Wearable Sensing and Machine Learning Methods" (2020). Graduate College Dissertations and Theses. 1303.
https://scholarworks.uvm.edu/graddis/1303