Estimating Sleep Stages using a Head Acceleration Sensor

Sensors

Motoki Yoshihi¹  Masaaki Makikawa²  Tianyi Wang²  Shima Okada²  Toshihiro Kitajima³ 

[1] Department of Robotics, Faculty of Science and Engineering, Ritsumeikan University Graduate Schools, Shiga 525-8577, Japan;Department of Robotics, Faculty of Science and Engineering, Ritsumeikan University, Shiga 525-8577, Japan;MD-4 Lab. Samsung R&D Institute Japan, Osaka 562-0036, Japan;

关键词: ballistocardiogram;    head acceleration sensor;    sleep stages;    sleep disruption;    REM sleep;   

DOI  :  10.3390/s21030952

来源: DOAJ

【 摘 要 】

Sleep disruption from causes, such as changes in lifestyle, stress from aging, family issues, or life pressures are a growing phenomenon that can lead to serious health problems. As such, sleep disorders need to be identified and addressed early on. In recent years, studies have investigated sleep patterns through body movement information collected by wristwatch-type devices or cameras. However, these methods capture only the individual’s awake and sleep states and lack sufficient information to identify specific sleep stages. The aim of this study was to use a 3-axis accelerometer attached to an individual’s head to capture information that can identify three specific sleep stages: rapid eye movement (REM) sleep, light sleep, and deep sleep. These stages are measured by heart rate features captured by a ballistocardiogram and body movement. The sleep experiment was conducted for two nights among eight healthy adult men. According to the leave-one-out cross-validation results, the F-scores were: awake 76.6%, REM sleep 52.7%, light sleep 78.2%, and deep sleep 67.8%. The accuracy was 74.6% for the four estimates. This proposed measurement system was able to estimate the sleep stages with high accuracy simply by using the acceleration in the individual’s head.

【 授权许可】

Unknown