AI-Driven Wearable Sensing Systems for Human Well-being

Wearable sensing technologies are evolving from simple activity trackers into intelligent, multimodal systems that can continuously assess and enhance human well-being. Traditional devices remain largely passive, limited by heterogeneous data, energy constraints, and opaque AI models. This thesis, titled AI-Driven Wearable Sensing Systems for Human Well-being, advances a new generation of wearable platforms that transform raw physiological and behavioral signals into real-time, adaptive, and interpretable decisions. Three interconnected domains are addressed. For mobility, multimodal systems combining textile EMG, IMU, and strain sensors enable robust gesture recognition, human activity monitoring, and physiologically informed exoskeleton control. For communication, wearable silent speech interfaces integrate throat strain and multi-channel EMG sensors with efficient neural decoders and language-model–assisted expansion, enabling natural, low-latency expression for both healthy users and individuals with dysarthria. For health monitoring, smart garments and multimodal home platforms provide continuous tracking of sleep, rehabilitation progress, and daily health states, moving beyond passive recording toward proactive, personalized guidance. These contributions converge in a future roadmap toward interpretable human body digital twins, where cross-scale sensing and adaptive AI together support trustworthy, personalized, and continuous health management. By uniting sensor innovation, embedded intelligence, and human-centered design, this thesis establishes a framework for AI-driven wearable systems that actively promote human well-being in both clinical and everyday contexts.

Name dc:type.qualificationname

Doctor of Philosophy (PhD)

Level dc:type.qualificationlevel

Doctoral

Grantor dc:publisher.institution

University of Cambridge

Year dc:date.issued

2025

Author dc:creator

• Tang, Chenyu

Advisor dc:contributor.advisor

• Occhipinti, Luigi

• Wearable Sensor
• Machine Learning
• Healthcare
• Human-Machine-Interaction
• Robotics

Licence

• https://www.repository.cam.ac.uk/bitstreams/71adbf77-d8dc-4b9e-bbea-5bf57262fd8a/download
• http://purl.org/NET/rdflicense/allrightsreserved

Language dc:language

eng

Author Identifier

0000-0002-6368-5639

OAI identifier oai:identifier

oai:www.repository.cam.ac.uk:1810/395965