Abstract Rationale Unobtrusive, long-term home monitoring of sleep is critical for understanding and managing sleep disorders. Current approaches rely on laboratory environment and expert scoring, which are resource-intensive, disruptive, and time limited. Upcoming wearable technologies may cause discomfort while sleeping, and rely on patient input (e.g. charging). Albus Home, a novel automated, contactless bedside monitor capable of simultaneously tracking sleep, respiratory rate and events interrupting sleep (such as cough), may support comprehensive diagnosis. While the monitor has multiple capabilities, this study focuses on the performance of Albus Home against a portable reference for two fundamental functions: 1) sleep/wake classification and 2) absence-from-bed detection. Methods We conducted a comparative validation study for both functions. For sleep/wake, overnight data from 6 adults (2 females; ages 28-64 years, BMI 20.8-37.1) for 3 nights each was recorded (total 18 nights). Cohort included single sleepers (n = 3) and co-sleepers (n = 3) with a mix of healthy subjects (n = 4), and those who reported respiratory conditions (asthma n = 1; sleep apnea n = 1). Simultaneous recordings were obtained from Albus Home and AASM-approved polygraphy (SOMNOtouch RESP, Somnomedics, Germany) as reference for 158.9 hours (mean 8.8h/night). Each 60-second epoch (total 9,534) was classified as asleep or awake automatically by each system’s proprietary software. For absence-from-bed detection, overnight data from 4 adults (1 female; 3 single sleepers, 1 co-sleeper) for 1 night each was simultaneously recorded using Albus Home and synchronized videography as reference for 40 hours (10h/night). 60-second epochs (total 2,400) were independently classified by humans from videography recording, and by software for Albus Home recording. For each function, epoch-based paradigm was followed to measure performance for each night against the reference. Results For sleep-wake classification, overnight per-epoch accuracy was 0.74 ± 0.13 (range 0.52-0.99), averaged across nights. Mean sensitivity to sleep was 0.90 (SD 0.20; range 0.80–0.99). Mean sensitivity to wake was 0.54 (SD 0.30; range 0.10–0.99). For absence-from-bed detection, overnight per epoch accuracy was 0.97, averaged across nights. Conclusions Preliminary validation of Albus Home against reference shows high sleep and absence-from-bed detection accuracy in real-world environments in a diverse cohort (co-sleepers, wide age and BMI range, respiratory disorders). These findings demonstrate Albus Home’s potential to objectively and accurately measure sleep parameters every night for as long as required without adding any patient burden and in their comfortable home environment. This unlocks unprecedented opportunities in healthcare and clinical trials for reliable daily sleep and nocturnal monitoring. This abstract is funded by: Albus Health (spinout company from University of Oxford)
Ozcan et al. (Fri,) studied this question.