Published: 31 October 2023| Version 1 | DOI: 10.17632/35dp3db8yj.1
zhi liu, Qinhan Zhang, Sixin Luo, Meiqiao Qin, Xu Lei


RichSleep dataset contains overnight PSG sleep records of 24 subjects and corresponding label files. The dataset captured a total of eighteen physiological data in the PSG recordings, including six EEG channels ('F3-A2', 'F4-A1', 'C3-A2', 'C4-A1', 'O1-A2', 'O2-A1'), two EOG channels('EOG-L', 'EOG-R'), chin EMG, ECG, respiratory signals, snoring, leg movement, blood oxygen and carbon dioxide levels, chest and abdominal movement, and breath interval duration. The *xn.edf files are whole-night polysmnographic sleep recordings.The *xn-ann.txt files contain annotations of the sleep patterns that correspond to the PSGs. These patterns (hypnograms) consist of sleep stages W, 1, 2, 3, R . To minimize the effect of impedance on physiological signals, impedance was kept below 10 ohms during the collection process, with a sampling frequency of 200 Hz.


Steps to reproduce

Data collection for the RichSleep dataset was supported by the Sleep and Neuroimaging Centre at Southwest University. Data collection was conducted in an isolated environment that is conducive to human sleep and rest. To mitigate external influences, windows were covered with opaque curtains to control lighting. Participants underwent physiological signal collection while resting on sleep beds. Prior to the experiment, volunteers were recruited from the local community, and after recruiting 37 volunteers, a screening process was carried out to select individuals who were physically healthy and free of any sleep-related disorders. A group of 12 males and 12 females, aged between 18 and 24 years, was finally selected to participate as subjects. Participants were also required to abstain from any medication for three days prior to the experiment. Each participant was allotted ten hours of sleep, from 10 pm to 8 am, over the course of 24 days.


Sleep Studies