Files and links (1)
CC BY-SA V4.0, Open Access
Dissertation
Wearable Systems and Machine Learning for Affect Recognition and Intervention
Washington State University
Doctor of Philosophy (PhD), Washington State University
05/2024
DOI:
https://doi.org/10.7273/000006477
Abstract
Stress recognition and monitoring from wearable sensor data is an emerging area of research with significant implications for an individual’s physical, social, and mental health. Mobile health interventions that incorporate real-time monitoring of physiological and behavioral markers of stress offer promise for delivering tailored interventions to individuals during high-risk states of heightened stress. This thesis presents a framework for stress detection using wearable sensor systems to facilitate real-time personalized intervention generation for stress management and regulation. We envision such a system to be incredibly beneficial for general as well as vulnerable population groups such as individuals diagnosed with alcohol use disorder. In a real-world study with individuals diagnosed with alcohol use disorder and undergoing rehabilitation programs to remain abstinent from alcohol, we studied the associations between alcohol craving/consumption, negative/positive mood, and stress. In doing so, we have studied, proposed, and validated solutions for important problems in mobile health sensor systems, such as sensor channel selection, optimal segment length, and processing of noisy sensor data. Furthermore, we present machine learning algorithms capable of learning from raw sensor data without feature computation and selection for stress detection and classification. We validated proposed machine learning algorithms on real-world and in-lab datasets. Finally, we examined the relationship between music listening and stress using data collected in a lab study. We also statistically established music listening as an intervention strategy for stress prevention and regulation. We lay the foundation of a framework for future research incorporating music listening for stress intervention in an intelligent mobile health system.
Our aim in this thesis is not only to validate the proposed pipeline for stress detection and intervention but also to solve important problems generally applicable to wearable sensor systems, examine associations between physiological and ecological markers of stress and alcohol addiction measures, discover acoustical properties of musical tracks correlated with stress, and motivate future research by making two stress datasets and other resources publicly available.
Metrics
4 File views/ downloads
49 Record Views
Details
- Title
- Wearable Systems and Machine Learning for Affect Recognition and Intervention
- Creators
- Ramesh Kumar Sah
- Contributors
- Diane Cook (Co-Chair) - Washington State University, School of Electrical Engineering and Computer ScienceHassan Ghasemzadeh (Co-Chair)Michael Cleveland (Committee Member)Ganapati Manjunath Bhat (Committee Member) - Washington State University, School of Electrical Engineering and Computer Science
- Awarding Institution
- Washington State University
- Academic Unit
- School of Electrical Engineering and Computer Science
- Theses and Dissertations
- Doctor of Philosophy (PhD), Washington State University
- Publisher
- Washington State University
- Number of pages
- 167
- Identifiers
- 99901120940201842
- Language
- English
- Resource Type
- Dissertation