Adaptive, Federated, and Resilient Health Monitoring via Novel Online Collaborative Learning Algorithms

Adaptive monitoring of a large population of dynamic processes is critical for the timely detection of abnormal events under limited resources. This issue is pervasive in various sectors, including healthcare and engineering systems, due to the disparity between available monitoring resources and the large population of units, and the uncertain and heterogeneous dynamics of unit progression. To effectively address this problem, in this dissertation, we introduce advanced methodologies for designing adaptive monitoring strategies. We first develop an online collaborative learning framework that efficiently models and monitors a population of dependent units under resource constraints. We then develop a decentralized online collaborative framework that enables online modeling and monitoring of units with latent dynamics while preserving data privacy. Finally, we develop a novel robust online multi-task learning algorithm designed to capture latent structures inherent in the population from sequentially observed data under corruption. We have demonstrated the effectiveness of the proposed methods through rigorously proven theoretical analysis and experiments, including simulation studies and real-world world applications including cognitive degradation monitoring in Alzheimer’s Disease (AD) and battery degradation monitoring.

Name thesis:degree_name

Doctor of Philosophy

Discipline thesis:degree_discipline

Industrial Engineering

Grantor

University of Houston

Year dc:date.issued

2025

Author dc:creator

• Kosolwattana, Tanapol 1996-

Advisor dc:contributor.advisor

• Lin, Ying

Committee members dc:contributor.committeemember

• Wang, Huazheng
• Hu, Renjie
• Feng, Qianmei
• Lim, Gino

• Industrial engineering

Language dc:language.iso

English

Handle dc:identifier.uri

https://hdl.handle.net/10657/19834

OAI identifier oai:identifier

oai:uh-ir.tdl.org:10657/19834