Mortality risk assessment using deep learning-based frequency analysis of electroencephalography and electrooculography in sleep
SLEEP · 10.02.2025 · 48 (2)
DOI: 10.1093/sleep/zsae219 ↗Kurz gefasst
Fachartikel in der Zeitschrift SLEEP, online erschienen am 20.09.2024. Verfasst von Teitur Óli Kristjánsson und 5 weiteren Autor:innen. Beteiligte Einrichtung: Department of Health Technology, Technical University of Denmark, Kongens Lyngby, Denmark. Hauptthemen laut MEDLINE: Deep Learning, Electroencephalography, Electrooculography, Mortality. Die Arbeit wurde bisher 5-mal zitiert (Crossref) und nennt 44 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: NHLBI NIH HHS, NCATS NIH HHS, NIA NIH HHS.
Abstract (Originalsprache)
Study Objectives To assess whether the frequency content of electroencephalography (EEG) and electrooculography (EOG) during nocturnal polysomnography (PSG) can predict all-cause mortality.
Methods
Power spectra from PSGs of 8716 participants, including from the MrOS Sleep Study and the Sleep Heart Health Study, were analyzed in deep learning-based survival models. The best-performing model was further examined using SHapley Additive Explanation (SHAP) for data-driven sleep-stage specific definitions of power bands, which were evaluated in predicting mortality using Cox Proportional Hazards models.
Results
Survival analyses, adjusted for known covariates, identified multiple EEG frequency bands across all sleep stages predicting all-cause mortality. For EEG, we found an all-cause mortality hazard ratio (HR) of 0.90 (CI: 95% 0.85 to 0.96) for 12–15 Hz in N2, 0.86 (CI: 95% 0.82 to 0.91) for 0.75–1.5 Hz in N3, and 0.87 (CI: 95% 0.83 to 0.92) for 14.75–33.5 Hz in rapid-eye-movement sleep. For EOG, we found several low-frequency effects including an all-cause mortality HR of 1.19 (CI: 95% 1.11 to 1.28) for 0.25 Hz in N3, 1.11 (CI: 95% 1.03 to 1.21) for 0.75 Hz in N1, and 1.11 (CI: 95% 1.03 to 1.20) for 1.25–1.75 Hz in wake. The gain in the concordance index (C-index) for all-cause mortality is minimal, with only a 0.24% increase: The best single mortality predictor was EEG N3 (0–0.5 Hz) with a C-index of 77.78% compared to 77.54% for confounders alone.
Conclusions
Spectral power features, possibly reflecting abnormal sleep microstructure, are associated with mortality risk. These findings add to a growing literature suggesting that sleep contains incipient predictors of health and mortality.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Health Technology, Technical University of Denmark, Kongens Lyngby, Denmark.
- Psychiatry and Behavioral Sciences, Stanford University, Palo Alto, CA, USA.
- Department of Clinical Neurophysiology, Rigshospitalet, Copenhagen, Denmark.
- Research Institute, California Pacific Medical Center, San Francisco, CA, USA.
- Department of Epidemiology and Biostatistics, University of California, San Francisco, CA, USA.
ORCID-Profile: Teitur Óli Kristjánsson, Andreas Brink-Kjaer, Emmanuel Mignot
Förderung
- NHLBI NIH HHS
- NCATS NIH HHS
- NIA NIH HHS
- NIAMS NIH HHS
Zitieren
APA
Kristjánsson, T. Ó., Stone, K. L., Sorensen, H. B. D., Brink-Kjaer, A., Mignot, E., & Jennum, P. (2025). Mortality risk assessment using deep learning-based frequency analysis of electroencephalography and electrooculography in sleep. SLEEP, 48(2). https://doi.org/10.1093/sleep/zsae219
Vancouver
Kristjánsson TÓ, Stone KL, Sorensen HBD, Brink-Kjaer A, Mignot E, Jennum P. Mortality risk assessment using deep learning-based frequency analysis of electroencephalography and electrooculography in sleep. SLEEP. 2025;48(2). doi:10.1093/sleep/zsae219
BibTeX
@article{Kristjnsson2025,
author = {Kristjánsson, Teitur Óli and Stone, Katie L and Sorensen, Helge B D and Brink-Kjaer, Andreas and Mignot, Emmanuel and Jennum, Poul},
title = {Mortality risk assessment using deep learning-based frequency analysis of electroencephalography and electrooculography in sleep},
journal = {SLEEP},
year = {2025},
volume = {48},
number = {2},
doi = {10.1093/sleep/zsae219}
}
EUVIMED