The Virtual Sleep Lab—A Novel Method for Accurate Four-Class Sleep Staging Using Heart-Rate Variability from Low-Cost Wearables DOI Creative Commons
Pavlos Topalidis, Dominik Philip Johannes Heib, Sebastian Baron

et al.

Sensors, Journal Year: 2023, Volume and Issue: 23(5), P. 2390 - 2390

Published: Feb. 21, 2023

Sleep staging based on polysomnography (PSG) performed by human experts is the de facto “gold standard” for objective measurement of sleep. PSG and manual sleep is, however, personnel-intensive time-consuming it thus impractical to monitor a person’s architecture over extended periods. Here, we present novel, low-cost, automatized, deep learning alternative that provides reliable epoch-by-epoch four-class approach (Wake, Light [N1 + N2], Deep, REM) solely inter-beat-interval (IBI) data. Having trained multi-resolution convolutional neural network (MCNN) IBIs 8898 full-night manually sleep-staged recordings, tested MCNN classification using two low-cost (<EUR 100) consumer wearables: an optical heart rate sensor (VS) breast belt (H10), both produced POLAR®. The overall accuracy reached levels comparable expert inter-rater reliability devices (VS: 81%, κ = 0.69; H10: 80.3%, 0.69). In addition, used H10 recorded daily ECG data from 49 participants with complaints course digital CBT-I-based training program implemented in App NUKKUAA™. As proof principle, classified extracted captured sleep-related changes. At end program, reported significant improvements subjective quality onset latency. Similarly, latency showed trend toward improvement. Weekly latency, wake time during sleep, total also correlated significantly reports. combination state-of-the-art machine suitable wearables allows continuous accurate monitoring naturalistic settings profound implications answering basic clinical research questions.

Language: Английский

SleepPPG-Net: A Deep Learning Algorithm for Robust Sleep Staging From Continuous Photoplethysmography DOI

Kevin Kotzen,

Peter Charlton, Sharon Salabi

et al.

IEEE Journal of Biomedical and Health Informatics, Journal Year: 2022, Volume and Issue: 27(2), P. 924 - 932

Published: Nov. 29, 2022

Sleep staging is an essential component in the diagnosis of sleep disorders and management health. traditionally measured a clinical setting requires labor-intensive labeling process. We hypothesize that it possible to perform automated robust 4-class using raw photoplethysmography (PPG) time series modern advances deep learning (DL). used two publicly available databases included PPG recordings, totalling 2,374 patients 23,055 hours continuous data. developed SleepPPG-Net, DL model for from series. SleepPPG-Net was trained end-to-end consists residual convolutional network automatic feature extraction temporal capture long-range contextual information. benchmarked performance against models based on best-reported state-of-the-art (SOTA) algorithms. When held-out test set, obtained median Cohen's Kappa ( $\kappa$ ) score 0.75 0.69 best SOTA approach. showed good generalization external database, obtaining 0.74 after transfer learning. Overall, provides new performance. In addition, high enough open path development wearables meet requirements usage applications such as monitoring obstructive apnea.

Language: Английский

Citations

28
The status and perspectives of nanostructured materials and fabrication processes for wearable piezoresistive sensors DOI Open Access
William Chiappim, Mariana Amorim Fraga, Humber Furlan

et al.

Microsystem Technologies, Journal Year: 2022, Volume and Issue: 28(7), P. 1561 - 1580

Published: March 17, 2022

Language: Английский

Citations

27
Certainty about uncertainty in sleep staging: a theoretical framework DOI Open Access
Hans van Gorp, Iris A. M. Huijben, Pedro Fonseca

et al.

SLEEP, Journal Year: 2022, Volume and Issue: 45(8)

Published: June 8, 2022

Abstract Sleep stage classification is an important tool for the diagnosis of sleep disorders. Because staging has such a high impact on clinical outcome, it that done reliably. However, known uncertainty exists in both expert scorers and automated models. On average, agreement between human only 82.6%. In this study, we provide theoretical framework to facilitate discussion further analyses staging. To end, introduce two variants uncertainty, from statistics machine learning community: aleatoric epistemic uncertainty. We discuss what these types uncertainties are, why distinction useful, where they arise staging, recommendations how can improve future.

Language: Английский

Citations

26
Recent Progress in Long-Term Sleep Monitoring Technology DOI Creative Commons

Jiaju Yin,

Jiandong Xu, Tian‐Ling Ren

et al.

Biosensors, Journal Year: 2023, Volume and Issue: 13(3), P. 395 - 395

Published: March 17, 2023

Sleep is an essential physiological activity, accounting for about one-third of our lives, which significantly impacts memory, mood, health, and children’s growth. Especially after the COVID-19 epidemic, sleep health issues have attracted more attention. In recent years, with development wearable electronic devices, there been studies, products, or solutions related to monitoring. Many mature technologies, such as polysomnography, applied clinical practice. However, it urgent develop non-contacting devices suitable household continuous This paper first introduces basic knowledge significance Then, according types signals monitored, this describes research progress bioelectrical signals, biomechanical biochemical used not ideal monitor quality whole night based on only one signal. Therefore, reviews multi-signal monitoring systematic schemes. Finally, a conclusion discussion are presented propose potential future directions prospects

Language: Английский

Citations

16
The Virtual Sleep Lab—A Novel Method for Accurate Four-Class Sleep Staging Using Heart-Rate Variability from Low-Cost Wearables DOI Creative Commons
Pavlos Topalidis, Dominik Philip Johannes Heib, Sebastian Baron

et al.

Sensors, Journal Year: 2023, Volume and Issue: 23(5), P. 2390 - 2390

Published: Feb. 21, 2023

Sleep staging based on polysomnography (PSG) performed by human experts is the de facto “gold standard” for objective measurement of sleep. PSG and manual sleep is, however, personnel-intensive time-consuming it thus impractical to monitor a person’s architecture over extended periods. Here, we present novel, low-cost, automatized, deep learning alternative that provides reliable epoch-by-epoch four-class approach (Wake, Light [N1 + N2], Deep, REM) solely inter-beat-interval (IBI) data. Having trained multi-resolution convolutional neural network (MCNN) IBIs 8898 full-night manually sleep-staged recordings, tested MCNN classification using two low-cost (<EUR 100) consumer wearables: an optical heart rate sensor (VS) breast belt (H10), both produced POLAR®. The overall accuracy reached levels comparable expert inter-rater reliability devices (VS: 81%, κ = 0.69; H10: 80.3%, 0.69). In addition, used H10 recorded daily ECG data from 49 participants with complaints course digital CBT-I-based training program implemented in App NUKKUAA™. As proof principle, classified extracted captured sleep-related changes. At end program, reported significant improvements subjective quality onset latency. Similarly, latency showed trend toward improvement. Weekly latency, wake time during sleep, total also correlated significantly reports. combination state-of-the-art machine suitable wearables allows continuous accurate monitoring naturalistic settings profound implications answering basic clinical research questions.

Language: Английский

Citations

14