0101 Performance of a Multisensor Ring to Evaluate Sleep At-Home Relative to PSG and Actigraphy: Importance of Generalized Versus Personalized Scoring

Abstract

Multisensor sleep wearable devices have demonstrated utility for research and relative accuracy for discerning sleep-wake patterns at home and in the laboratory. Additional sensors and more complex scoring algorithms may improve the ability of wearables to assess sleep health. Thirty-six healthy adults completed assessment while wearing the experimental device (Happy Ring), as well as Philips Actiwatch, Fitbit, Oura, and Whoop devices. Evaluations at home were conducted using the Dreem headband as an at-home polysomnography reference. The experimental Happy Ring device includes accelerometry, photoplethysmography, electrodermal activity, and skin temperature. Epoch-by-epoch analyses compared the Happy Ring to home polysomnography, as well as other sleep-tracking wearable devices. Scoring was accomplished using two machine-learning-derived algorithms: a “generalized” algorithm, similar to that used in other devices, which was static and applied to all users, and a “personalized” algorithm where parameters are personalized, dynamic, and change based on data collected across different parts of the night of sleep. Compared to home polysomnography, the Happy generalized algorithm demonstrated good sensitivity (94%) and specificity (67%), and the Happy personalized algorithm also performed well (93% and 75%, respectively). Other devices demonstrated good sensitivity, ranging from 91% (Whoop) to 96% (Oura). However, specificity was more variable, ranging from 41% (Actiwatch) to 60% (Fitbit). Overall accuracy using the Happy Ring was 91% for generalized and 92% for personalized algorithms, compared to 92% for Oura, 89% for Whoop, 89% for Fitbit, and 89% for Actiwatch. Regarding sleep stages, accuracy for the Happy Ring was 66%, 83%, and 78% for light, deep, and REM sleep, respectively, for the generalized algorithm. For the personalized algorithm accuracy was 78%, 92%, and 95%, for light, deep and REM sleep, respectively. Post-hoc analyses showed that the Happy personalized algorithm demonstrated better specificity than all other modalities (p<0.001). Kappa scores were 0.42 for generalized and 0.60 for personalized, compared to 0.45 for Oura, 0.47 for Whoop, and 0.48 for Fitbit. The multisensory Happy ring demonstrated good sensitivity and specificity for the detection of sleep at home. The personalized approach outperformed all others, representing a potential innovation for improving detection accuracy. Dr. Grandner is supported by R01DA051321 and R01MD011600. This work was supported by Happy Health, Inc.