WristTracker : Smartwatch-based mobile healthcare to track unconscious wrist movements = 기계학습 알고리즘을 활용한 스마트워치 기반의 움직임 트래킹 모바일 헬스케어 시스템 설계
Smartwatch-based mobile healthcare to track unconscious wrist movements
Understanding one's unconscious movements during sleep is very important in various healthcare monitoring applications. Since sleep behaviors have significant impact on sleep quality and further quality of life, spatial information of wrist movements may complement a limited pathological feedback and help management of disorders. In this work, we propose WristTracker -- a smart-watch based system to track wrist movements during sleep to understand unconscious behavior. Possible application scenario is identifying scratch locations for patients with various aspects of pruritus is a source to determine if a medication had an effect or not. Another important application is understanding sleep behavior or posture, in particular, for patients with sleep disorder such as sleep apnea which are associated with sleep posture. To attain this end, we provide preliminary results on how WristTracker can be used to unobtrusively recognize scratch location with three body segments as well as sleep position with four postures. Features of obtained data are extracted from inertial sensor on smartwatch, and are used as training data. Cross-validation methods are applied to data with six patients in home environment, and thereby we verify the tracking system.