Files and links (1)
CC BY V4.0, Open Access
Dissertation
Creating Interpretable Data-Driven Approaches for Remote Health Monitoring
Washington State University
Doctor of Philosophy (PhD), Washington State University
2023
DOI:
https://doi.org/10.7273/000005346
Abstract
The impending "age wave'' underscores the urgent need for technology that supports remote health monitoring. This research focuses on developing technologies to improve clinician interpretability of remote health monitoring systems. These technologies visualize and summarize raw data and machine learning inferences. Unlike prior methods, this dissertation describes techniques developed in partnership with the target end user: clinicians. Keeping the clinician in the design loop improves technology trust, adoption, and patient outcomes.
In this research, we introduce two visualization studies that illustrate and validate the human-in-the-loop approach, demonstrating that with this technique clinicians can successfully identify health events from unconventional data sources, such as smart homes and smartwatches. To complement these studies, the research also introduce methods to create machine learning models that are interpretable for humans. Here, we focus on discovering and communicating patterns that summarize complex data and justify machine learning decisions. All of the introduced algorithms yield improved objective performance on synthetic and real-world complex data. To evaluate their efficacy for remote health monitoring, the methods are assessed by clinician end users in a series of comparative surveys. In these experiments, the methods introduced in this dissertation show significantly enhanced interpretability. We discuss the implications and limitations of this work for remote health monitoring systems and offer ideas for continued work.
Metrics
1 File views/ downloads
21 Record Views
Details
- Title
- Creating Interpretable Data-Driven Approaches for Remote Health Monitoring
- Creators
- Alireza Ghods
- Contributors
- Diane Cook (Advisor)Trong Nghia Hoang (Advisor)Venkata Janardhan Rao Doppa (Committee Member)Hassan Ghasemzadeh (Committee Member)
- Awarding Institution
- Washington State University
- Academic Unit
- Electrical Engineering and Computer Science, School of
- Theses and Dissertations
- Doctor of Philosophy (PhD), Washington State University
- Publisher
- Washington State University
- Number of pages
- 268
- Identifiers
- 99901031138501842
- Language
- English
- Resource Type
- Dissertation