Approved Research

Development of an algorithm to investigate the landscape of human sleep in the real world

Lay summary

Sleep is essential for our quality of life. Moreover, there are known relations between sleep disorders and psychiatric disorders (e.g., depression) or neurodegenerative diseases (e.g., Parkinson's disease). Nowadays, wearable fitness devices such as the Fitbit or AppleWath measure activity and report users' sleep. However, these devices are not usable for a serious medical diagnosis because there is still not enough data to delineate the disease-characteristic sleep profiles distinctive to healthy profiles. We focus on wearable devices that measure acceleration (e.g., arm movements) and develop an accurate algorithm to predict sleep profiles. We will improve the algorithm by assessing the dataset stored in the U.K. BioBank, which is the world's largest dataset of accelerometers today. This study enhances our project to collect sleep data from a larger population, including healthy and unhealthy people. The big data resulting from the project will provide the landscape of human sleep and realizes the medical diagnosis based on the data recorded with wearable devices.