Improving Accuracy of Real-Time Sleep Stage Estimation by Considering Personal Sleep Feature and Rapid Change of Sleep Behavior

摘要

This paper proposes an improved method to estimate the sleep stage in real-time by heat rate obtained with a noncontact biological sensor device and investigates its effectiveness. The current version of real-time sleep stage estimation decreases its estimation accuracy in early term of sleeping and cannot follow rapid change of the sleep stage. To improve the estimation accuracy of the improved method, this paper introduces sleep feature of different day into the estimation model as user's pre-information and introduces WAKE and REM sleep stage heuristic classifiers.