Sleep monitoring with the Apple Watch: comparison to a clinically validated actigraph

Abstract

Background: We investigate the feasibility of using an Apple Watch for sleep monitoring by comparing its performance to the clinically validated Philips Actiwatch Spectrum Pro (the gold standard in this study), under free-living conditions. Methods: We recorded 27 nights of sleep from 14 healthy adults (9 male, 5 female). We extracted activity counts from the Actiwatch and classified 15-second epochs into sleep/wake using the Actiware Software. We extracted triaxial acceleration data (at 50 Hz) from the Apple Watch, calculated Euclidean norm minus one (ENMO) for the same epochs, and classified them using a similar algorithm. We used a range of analyses, including Bland-Altman plots and linear correlation, to visualize and assess the agreement between Actiwatch and Apple Watch. Results: The Apple Watch had high overall accuracy (97%) and sensitivity (99%) in detecting actigraphy-defined sleep, and adequate specificity (79%) in detecting actigraphy defined wakefulness. Over the 27 nights, total sleep time was strongly linearly correlated between the two devices (r=0.85). On average, the Apple Watch over-estimated total sleep time by 6.31 minutes and under-estimated Wake After Sleep Onset by 5.74 minutes. The performance of the Apple Watch compares favorably to the clinically validated Actiwatch in a normal environment. Conclusions: This study suggests that the Apple Watch could be an acceptable alternative to the Philips Actiwatch for sleep monitoring, paving the way for larger-scale sleep studies using Apple’s consumer-grade mobile device and publicly available sleep classification algorithms. Further study is needed to assess longer-term performance in natural conditions, and against polysomnography in clinical settings.