Work

Challenges and Opportunities in Wearable-Based Remote Patient Monitoring: Case Studies on Respiratory Diseases

Year: 2026
Type: dissertation
Abstract: Advancements in wearable sensing and remote patient monitoring have opened new avenues for managing chronic conditions. This dissertation examines how identifying and mitigating confounders that affect data quality in real-world settings enables the development of remote monitoring systems that yield more dependable and interpretable patient data. Through three works, it examines key issues related to data quality, patient engagement, and signal processing in real-world deployments. First, the dissertation reflects upon lessons learned from WearCOPD, a longitudinal study designed to detect COPD exacerbations using smartwatch-based sensing. Despite careful planning, post-study analysis revealed critical pitfalls, including participant adherence issues and unvalidated sensor data. Addressing these challenges in a second iteration of the study demonstrates strategies that enhance data reliability and improve study outcomes. Second, the dissertation addresses the challenge of extraneous voices and background noise in continuous speech monitoring for the detection of COPD exacerbations. The work introduces a novel primary speaker identification technique that distinguishes the primary speaker from others using daily speech characteristics captured by a worn microphone. Additionally, we contribute a novel speech feature extraction approach that distinguishes noise-resistant features from those that benefit from noise reduction, and from compromised features. Evaluating these methods on real-world patient data shows improvements in predictive performance, contributing to more robust speech-based health monitoring. Finally, the role of asynchronous messaging in remote patient monitoring is investigated through COVIDFree@Home, a platform deployed during the COVID-19 pandemic. The findings suggest that bidirectional patient-clinician communication fosters engagement, improves data accuracy, and enhances predictive modeling for patient deterioration. By integrating insights from these studies, this dissertation provides actionable recommendations for designing effective remote health monitoring systems. The findings emphasize the importance of iterative study design, patient engagement mechanisms, and robust signal processing techniques to maximize the utility of wearable sensing in healthcare.
Language: English
Open Access status: green
OpenAlex ID: W7204016472
Cites: 163
Cited by:
Related to: 9