Imputing missing sleep data from wearables with neural networks in real-world settings
SLEEP · 2024
DOI: 10.1093/sleep/zsad266 ↗Kurz gefasst
Fachartikel in der Zeitschrift SLEEP, 2024. Verfasst von Minki P Lee und 7 weiteren Autor:innen. Beteiligte Einrichtung: Department of Mathematics, University of Michigan, Ann Arbor, MI, USA. Hauptthemen laut MEDLINE: Sleep Disorders, Circadian Rhythm, Wearable Electronic Devices. Die Arbeit wurde bisher 13-mal zitiert (Crossref) und nennt 56 Literaturangaben. Gefördert von: Institute for Basic Sciences, Korean National Research Foundation, Korean Sleep Research Society.
Abstract (Originalsprache)
Sleep is a critical component of health and well-being but collecting and analyzing accurate longitudinal sleep data can be challenging, especially outside of laboratory settings. We propose a simple neural network model titled SOMNI (Sleep data restOration using Machine learning and Non-negative matrix factorIzation [NMF]) for imputing missing rest-activity data from actigraphy, which can enable clinicians to better handle missing data and monitor sleep–wake cycles of individuals with highly irregular sleep–wake patterns. The model consists of two hidden layers and uses NMF to capture hidden longitudinal sleep–wake patterns of individuals with disturbed sleep–wake cycles. Based on this, we develop two approaches: the individual approach imputes missing data based on the data from only one participant, while the global approach imputes missing data based on the data across multiple participants. Our models are tested with shift and non-shift workers' data from three independent hospitals. Both approaches can accurately impute missing data up to 24 hours of long dataset (>50 days) even for shift workers with extremely irregular sleep–wake patterns (AUC > 0.86). On the other hand, for short dataset (~15 days), only the global model is accurate (AUC > 0.77). Our approach can be used to help clinicians monitor sleep–wake cycles of patients with sleep disorders outside of laboratory settings without relying on sleep diaries, ultimately improving sleep health outcomes.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Mathematics, University of Michigan, Ann Arbor, MI, USA.
- Institute of Mathematics, EPFL, Lausanne, Switzerland.
- Department of Artificial Intelligence Convergence, Kangwon National University, Chuncheon, Republic of Korea.
- Department of Mathematical Sciences, KAIST, Daejeon, Republic of Korea.
- Biomedical Mathematics Group, Institute for Basic Science, Daejeon, Republic of Korea.
- Department of Neurology, Neuroscience Center, Samsung Biomedical Research Institute, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Republic of Korea.
ORCID-Profile: Sungkyu Park, Won Chang, Jae Kyoung Kim
Förderung
- Institute for Basic Sciences
- Korean National Research Foundation
- Korean Sleep Research Society
- University of Cincinnati Taft Research Center
- National Research Foundation of Korea
- Ministry of Education
Zitieren
APA
Lee, M. P., Hoang, K., Park, S., Song, Y. M., Joo, E. Y., Chang, W., Kim, J. H., & Kim, J. K. (2024). Imputing missing sleep data from wearables with neural networks in real-world settings. SLEEP. https://doi.org/10.1093/sleep/zsad266
Vancouver
Lee MP, Hoang K, Park S, Song YM, Joo EY, Chang W, et al. Imputing missing sleep data from wearables with neural networks in real-world settings. SLEEP. 2024. doi:10.1093/sleep/zsad266
BibTeX
@article{Lee2024,
author = {Lee, Minki P and Hoang, Kien and Park, Sungkyu and Song, Yun Min and Joo, Eun Yeon and Chang, Won and Kim, Jee Hyun and Kim, Jae Kyoung},
title = {Imputing missing sleep data from wearables with neural networks in real-world settings},
journal = {SLEEP},
year = {2024},
doi = {10.1093/sleep/zsad266}
}
EUVIMED