Enabling Closed-Loop Personalized Sleep Care through High-Fidelity Brain Tracking and Just-in-Time Brain Stimulation Wearables

Abstract

Generated by the functioning of many bodily sources, physiological signals hold essential information about the state of the body's major structures, including brain activities, eye movements, muscle contractions, and cardio-respiratory features, to name a few. Therefore, monitoring and stimulating such biosignals widely help diagnose, treat, and prevent several health conditions and diseases. This dissertation introduces novel sensing and stimulation capabilities for physiological signals focusing on mobile sleep health. Specifically, we present wearable systems and algorithms to provide clinical-grade solutions to concurrently sense multiple biosignals at non-standard locations and perform real-time brain entrainment for closed-loop personalized sleep care practices. Eventually, we deliver three fundamental contributions. First, we build practical monitoring systems comfortably placed either inside the ear canals or around the head for continuous use and that acquire all physiological signals of interest to sleep quality assessment. Second, we develop automatic techniques to separate biopotential individuals from the mixed head-based physiological signals to represent clean bioelectrical activities. Last, we establish smart in-time auditory stimulation to enable a complete closed-loop personalized sleep aid system for effective sleep assistance. To deliver these contributions, we exploit characteristics of the physiological signals, analyze and understand how they conceptually define sleep stages, work across different limitations and hardware-software barriers, and introduce novel wearables and dedicated algorithms to address challenges. We formulate, implement, and evaluate the proposed systems on real human subjects across different study protocols and demonstrate how they can enable more diverse real-world healthcare directions.