Real world sleep tracking in home-settings, is a cornerstone for personalized wellness monitoring. This helps in customizing diet, lifestyle and exercise1.
Poor sleep, over time has dramatic impact on cognitive, endocrine and metabolic functions of the body. 2,3,4
Wrist vs finger worn wearables are the two main ways of sleep monitoring in such conditions. However clinical studies report performance of wearables individually against sleep testing gold standard- polysomnography (PSG). PSG requires controlled settings in a sleep lab to capture sleep brain activity accurately.
So how do wrist and finger-based sleep tracking predict gross features of sleep-arousal (awake) periods for an individual in real-world settings?
Methods
7 volunteers within the Ultrahuman team; age range 25-45 years; mix of males(5) and females (2). Participants wore the Ultrahuman Ring AIR ® and the Apple watch models (various models) on the same hand/arm and acquired a minimum of 4 nights of continuous data and a total of 6 nights in all.
Volunteers did not have any changes in working hours, diet or supplements/medicines taken during the study period, with comparable stimulant intake. Statistical analyses of regression analyses used LMM or GLMM as per dataset characteristics.
Initial regression analyses using Pearson’s correlation coefficients were also undertaken.
Results
Pearson's correlation coefficient indicating overlap of sleep metrics between Ultrahuman RingAir and Apple Watch.
Gross measures of sleep duration are key metrics reported by all fitness trackers. These are used to estimate sleep and awake times. Given the empirical relationship between sleep quality and fragmentation on cognitive and physical function the following day, alignment between different tracking methods (wrist vs finger) is required. We found a high overlap between Apple Watch and Ultrahuman Ring AIR on four sleep metrics with initial Pearson’s coefficient of 0.94, 0.93 and 0.89 for sleep start, sleep end and total sleep duration respectively. LMM based regression results are reported in Table 1.
Table 1: Linear mixed model regression results
These coefficients estimate the difference in sleep metrics time between the Ultrahuman Ring AIR and the Apple Watch. There was no significant difference in the readouts of the two devices tested, indicating the equivalence of the Apple Watch with Ultrahuman Ring AIR on tracking these parameters.
In simplified terms, the two devices differed by 5.5 and 7.4 min in identifying leep start and end, while the difference in sleep duration calculation is 0.3 mins, with Apple watch recording a slightly longer sleep session. Ultrahuman Ring AIR records awake episodes longer by 6.5 minutes.
Conclusion
This pilot study aimed to compare a globally-used, consumer wrist-based wearable with the Ultrahuman Ring AIR to examine whether wrist- and finger-based sleep tracking predict gross features of sleep-arousal (awake) periods for an individual in real-world settings. Performance of the Ultrahuman Ring AIR is within the ranges of other wrist and finger-based sleep trackers5. Therefore use can be envisaged in cases where users want to determine how frequently they wake up in the night, and when they went to sleep or woke up. These results echo the findings of Chinoy et al., 2022 6, where comparison of four different types of consumer wearables was conducted and found that sleep-tracking performance was largely similar in unrestricted home sleep conditions.
This report is an initial study and had a small sample set and no gold standard reference device to comparison. The study was motivated by inquiries of users of both devices on overlaps and divergence of sleep measurements in the real world. It is pertinent to note that neither devices carry medical-grade approval from FDA or a similar organisation, hence diagnosis of sleep disorders is best done under the supervision of certified professionals using specialised equipment.
Reach out to support@ultrahuman.com for commercial queries and science@ultrahuman.com for scientific queries.
