Abstract

Sleep is a (generally) unobserved state, making it difficult to characterize normal sleeping patterns, as well as sleep disorders and their relationship to daytime function. The ubiquity of the internet of things provides a unique opportunity to gather a more realistic picture of sleep through continuous monitoring. However, most commercial devices lack sufficient validation of at-home deployment, and often require some degree of physical contact to gather signals. We sought to assess a contactless, computer-vision sleep system’s ability to detect sleep-disordered breathing events. This first phase of model development derived breathing activity from reference points tracked by the computer vision system. Simultaneous comparison was made to a NoxT3 level 3 home sleep apnea testing device. Signals from the camera algorithm were correlated with the respiratory excursion signals from the respiratory inductance plethysmogram. Expert scorers were blinded to subject identity on randomly and independently presented algorithm-derived breathing signals as well as Nox studies. Breathing disturbances defined on the algorithm-derived signal and AASM-defined apneas and hypopneas from the NoxT3 study were compared. Comparison between the computer-vision breathing signal and NoxT3 abdominal RIP band demonstrated high fidelity, rho=0.921 (p<0.001). On 7 pilot participants with minimal sleep-disordered breathing (AHI 10.3 ± 4.6), diagnostic accuracy was high (rho=0.77, p=0.043). A Bland-Altman plot was used to assess bias in scoring on the computer-vision-derived breathing signal, indicating that there was a slight tendency to underscore events (bias of -4.7) when additional oxygenation and heart rate signals were not present. This study demonstrates that a computer-vision-based platform provides sufficiently high accuracy to contactlessly monitor sleep breathing disturbances on a continuous basis. This offers the potential to provide individuals low-cost screening and/or baseline sleep evaluations, with the ability to automatically spotlight abnormalities, reducing the burden of collecting such large volumes of longitudinal data. Moreover, such longitudinal data can allow for more robust estimates of true sleep-breathing variability and response to therapies in a real-world environment. None

Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call

Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.