In recent years, many wearable devices, e.g, wristbands and smartwatches, provide sleep monitoring function. Those sleep annotations are usually based on features extracted from the accelerometer signal (to detect body movement), which could reliably distinguish wake from sleep under most conditions, but it is unlikely to differentiate "light" from "deep" sleep. Some more advanced models of wearable devices incorporate heart rate patterns, measured by photoplethysmogram (PPG) signal, to distinguish the subtle differences in different sleep stages ("light", "deep", and REM).
Our group is involved in several epidemiologic studies (see for example: the Explore TCM project, and the WELL-China project) that need to gather long-term behavioral patterns. The wristband is user-friendly and low-cost, and could provide long-term exercise/activity information, basic sleep structure and objective sleep quality assessment. Thus, it might be a valuable solution for those research projects.
Therefore, we compared the sleep annotation by a class of wristbands/smartwatches that implemented an algorithm that is based on both body movement and heart rate variability (HRV). The analysis results were compared to an ECG-based algorithm for sleep analysis developed by BIDMC researchers (Thomas, Mietus, Peng, and Goldberger), called Cardiopulmonary Coupling (CPC) analysis.
The following wearable devices have been studied (click the link for the test results):
- HUAWEI FIT
- HUAWEI Band 2 Pro
- HUAWEI Watch 2 and Watch 2 Pro
- HUAWEI Talkband B5
- HUAWEI Band 3 Pro
- HUAWEI Watch GT
- honor Band 4
- honor Watch
- Detailed reports for all devices