Brain Sleep Inc. (Headquarters: Chiyoda-ku, Tokyo, CEO: Atsushi Hirota, hereinafter referred to as "Brain Sleep"), in collaboration with Professor Emeritus Keiki Takadama of the University of Electro-Communications (hereinafter referred to as "UEC") (currently Professor at the Interdisciplinary Information Science Research Division, Information Infrastructure Center, The University of Tokyo), conducted research last year with the aim of improving the accuracy of sleep stage estimation in the sleep measurement tool "Brain Sleep Coin," and developed a completely new concept sleep stage estimation algorithm that can grasp "sleep rhythms." Brain Sleep will release an update to the sleep measurement tool "Brain Sleep Coin" equipped with this algorithm on Wednesday, September 25, 2024. background Sleep is one of the physiological phenomena that is important for health and happiness. Sleep problems and feelings of deep sleep are often subjective evaluations and may differ from objective sleep data, so it is important to understand sleep quality from both the main complaints and objective sleep data. The conventional method for measuring sleep quality is a method called polysomnography (PSG), which is manually evaluated by an expert. However, this method requires attaching many electrodes to the head and face, which places a significant physical and mental burden on the patient. Recently, with the advancement of AI and IT technology, sleep measurement methods using sensors that can be used in daily life without burden have attracted attention, and among them, methods using accelerometers are used in many research studies because they do not require a special environment. However, with conventional methods using accelerometers, there is no established method for accurately measuring sleep quality, and the accuracy is unclear. In particular, many sleep stage estimation algorithms using accelerometers do not use heart rate data, making it difficult to determine REM sleep or deep sleep. Brain Sleep has developed and sold the sleep measurement application "Brain Sleep Coin (app)" and the sleep measurement device "Brain Sleep Coin (device)" since October 2022 as tools for easily measuring sleep at home. This time, we conducted joint research with the University of Electro-Communications with the aim of further improving the accuracy of the sleep stage estimation of "Brain Sleep Coin". As a result, we have established a new method of measuring sleep quality based on body movements and a person's biological rhythms during sleep using an acceleration sensor attached to the waist. This is a method based on a completely different new concept that distinguishes it from conventional sleep stage estimation methods that focus only on accuracy. Research Summary and Results We discovered a correlation between body movement patterns obtained by an acceleration sensor and sleep stages (awake, REM sleep, non-REM sleep stage 1-2 (hereinafter referred to as "NREM12"), and non-REM sleep stage 3 (hereinafter referred to as "NREM3")), and based on that we developed a sleep stage estimation algorithm and verified its effectiveness. Specifically, 35 subjects in their 20s to 50s were asked to wear the Brain Sleep Coin (device) and PSG tests were conducted to collect and analyze the data. Noting the correlation between the frequency of body movements obtained from the Brain Sleep Coin (device) worn on the waist and the ultradian rhythm, one of the biological rhythms during human sleep, the researchers succeeded in estimating the ultradian rhythm from body movements. Based on the estimated rhythm, they developed an algorithm that can estimate REM sleep and NREM3, which are difficult to estimate from body movements alone. ●Focusing on the accuracy of NREM3 estimation, experiments have confirmed that it can estimate with higher accuracy than conventional machine learning models. Specifically, the precision rate (the percentage of parts correctly estimated as NREM3) improved by 27%, and the recall rate (the percentage of parts that were correctly estimated as NREM3) improved by 24%. As a result, the overall matching rate improved by 26% compared to the conventional Brain Sleep Coin sleep stage judgment algorithm. ●The figure compares sleep stages from PSG testing with sleep stage estimation results from the new algorithm. The vertical axis represents sleep stages, and the horizontal axis represents time. The blue line represents the correct sleep stages obtained from PSG testing, and the orange line represents the sleep stages estimated by the algorithm. Circles indicate areas where NREM3 was correctly estimated, and crosses indicate areas where it was incorrectly estimated. The new algorithm has significantly reduced misestimation and has succeeded in capturing sleep rhythms. This joint research has made it possible to provide simpler and more accurate sleep monitoring. It is expected that this will help many people understand their sleep conditions and get better sleep. The details of this research have been announced as follows.・NREM3 sleep: NREM3 Sleep Stage Estimation Based on Accelerometer by Body Movement Count and Biological Rhythms, AAAI 2024 Spring Symposium Series・REM sleep: REM Estimation Based on Accelerometer by Excluding Other Stages and Two-Scale Smoothing, IEEE EMBC 2024 Comment from co-researcher (Professor Emeritus Takadama, University of Electro-Communications) There are many products that boast high accuracy, but from a health perspective, it is more important to understand the sleep rhythm in which deep sleep and light sleep alternate. Achieving accurate estimations is a challenge, but improvements in REM sleep and NREM3 estimations have not only improved overall accuracy but also given us a better understanding of our sleep rhythms. What is the sleep measurement tool "Brain Sleep Coin"? This sleep measurement tool is equipped with a unique measurement algorithm developed based on data accumulated through sleep research, and values sleep. It allows for more personalized analysis according to people's movements. It supports improving the quality of sleep by suggesting a virtuous cycle in which sleep scores are converted into coins through improving sleep habits. In addition to improving sleep stage estimation, this update also adds a function to summarize daily sleep data and visualize weekly, monthly, and yearly sleep data in a radar chart. A new function has also been added that analyzes days when users reported they were in good condition and their sleep data to suggest optimal sleep times and actions for users. Product details page: https://brain-sleep.com/collections/sleep-tech-app/products/coin Name of University: University of Electro-Communications, National University Corporation Address: 1-5-1 Chofu-gaoka, Chofu-shi, Tokyo 182-8585 President: Shunichi Tano University website: https://www.uec.ac.jp/ University Overview: Originating from the Wireless Telegraph Training School of the Telegraph Association, which was established in 1918 with the aim of training wireless communication engineers, the university opened in 1949 under the name "University of Electro-Communications." It is the only national university with faculties that does not include the name of a place in its name, which is based on the spirit of creating a university that is open to the whole of Japan. Currently, it is a national university of science and technology that provides education and research in a wide range of fields from the basics to applications of science and engineering, including not only information, electricity, and communications, but also physical engineering, materials science, life science, optical science, electronics, robotics, mechanical engineering, and media.
Log in
