0314 PSG Validation Of Minute-to-minute Scoring For Sleep And Wake Periods In A Consumer Wearable Device. (27th April 2018)

• Record Type: Journal Article
• Authors: Cheung, J; Zeitzer, J; Lu, H; Leary, E; Mignot, E
• Is Part Of: Sleep. Volume 41(2018)Supplement 1
• Page Start: A120
• Page End: A121
• Publication Date: 2018-04-27
• Subjects: Sleep -- Physiological aspects -- Periodicals; Sleep disorders -- Periodicals; Sommeil -- Aspect physiologique -- Périodiques; Sommeil, Troubles du -- Périodiques; Sleep disorders; Sleep -- Physiological aspects; Sleep -- physiological aspects; Sleep Wake Disorders; Psychophysiology; Electronic journals; Periodicals; 616.8498
• DOI: 10.1093/sleep/zsy061.313
• Languages: English
• ISSNs: 0161-8105

Abstract

Abstract: Introduction: Actigraphs are portable wrist-worn devices that record tri-axial accelerometry that are used both clinically and in research studies. The expense of research-grade actigraphs, however, limit their wide-spread adoption, especially in clinical settings. Tri-axial accelerometer-based consumer wearable devices have gained worldwide popularity and hold potential for a cost-effective alternative to the more expensive devices used in sleep research. The lack of independent validation of minute-to-minute accelerometer data for these consumer wearable devices has hindered their utility and acceptance. Methods: We studied a consumer-grade wearable device, Arc ($50, Huami Inc., Mountain View CA) for which minute-to-minute tri-axial accelerometer data (vector magnitude) were made assessible. Seventeen sleep clinic patients, including patients diagnosed with OSA, undergoing PSG wore on their non-dominant wrist both an Arc device and a research-grade actigraph (Actiwatch Spectrum, Philips, Bend OR). Time-stamped minute-to-minute data from each participant were aligned with scored PSG studies. The Cole-Kripke algorithm was used to determine sleep or wake for each 60s epoch on both the Arc and Actiwatch. Results: Compared to the gold-standard PSG, Arc has an accuracy of 89.5 ± 1.1% (SEM), sleep sensitivity (or wake specificity) was 95.3 ± 1.1%, and sleep specificity (wake sensitivity) was 50.1 ± 5.0%; while Actiwatch has an accuracy of 89.9 ± 1.1% (SEM), sleep sensitivity (or wake specificity) was 95.5 ± 0.9%, and sleep specificity (wake sensitivity) was 52.0 ± 4.2%. Conclusion: Preliminary results indicate that, as compared to minute-to-minute PSG, sleep and wake estimates generated by a consumer-grade wearable (Arc) were comparable to those generated by a clinical-grade actigraph (Actiwatch Spectrum). Further studies may allow for optimization of sleep/wake algorithm to improve accuracy of Arc. Support (If Any): Stanford CTSA UL1 TR001085, Klarman Family Foundation, NIH T32 HL110952, NIH K23 NS101094.