ChatPPG.
Join free
First page or project page of “Deep learning for obstructive sleep apnea diagnosis based on single channel oximetry” (opens in a new tab)

Paperdoi:10.1038/s41467-023-40604-3

Deep learning for obstructive sleep apnea diagnosis based on single channel oximetry

Venue listed2023Added Oct 29, 2026

  • JL
  • DA
  • FC
  • JV

Authors: Jeremy Levy, Daniel Álvarez, Félix Del Campo, Joachim A. Behar (Technion AIMLab / Valladolid)

Nature Communications 14, 2023

Publisher (opens in a new tab) Cite
First page or project page of “Deep learning for obstructive sleep apnea diagnosis based on single channel oximetry” (opens in a new tab)

TL;DR

OxiNet estimates the apnea-hypopnea index from single-channel oximetry, developed and tested on 12,923 polysomnography recordings from six databases. Trained on SHHS, it reached ICC 0.94 on external MESA and MrOS and 0.92 on CFS, missing 0.2% of moderate-to-severe patients, with lower performance in Black participants and in COPD.

  • Form factor: finger
Caveats
  • Trained on 90% SHHS1; MESA fully external; no PPG waveform.
  • Same cohort as training; SHHS2 0.95.
  • External.
  • External; lower in Black participants.
  • Severity from estimated AHI.
  • Same cohort as training; SHHS2 0.83.

Abstract

Obstructive sleep apnea (OSA) is a serious medical condition with a high prevalence, although diagnosis remains a challenge. Existing home sleep tests may provide acceptable diagnosis performance but have shown several limitations. In this retrospective study, we used 12,923 polysomnography recordings from six independent databases to develop and evaluate a deep learning model, called OxiNet, for the estimation of the apnea-hypopnea index from the oximetry signal. We evaluated OxiNet performance across ethnicity, age, sex, and comorbidity. OxiNet missed 0.2% of all test set moderate-to-severe OSA patients against 21% for the best benchmark.

Verbatim from the publisher's DOI record (opens in a new tab) · Publisher licence: https://creativecommons.org/licenses/by/4.0

Results on ChatPPG leaderboards · 8

All leaderboards
LeaderboardMethodResultRank
CFS ICCSleep apnea (OSA) screeningOxiNetCaveat: External; lower in Black participants.0.92ICC, higher is better#1 of 1
CFS macro-F1Sleep apnea (OSA) screeningOxiNetCaveat: External.0.78macro-F1, higher is better#1 of 1
MESA ICCSleep apnea (OSA) screeningOxiNetCaveat: Trained on 90% SHHS1; MESA fully external; no PPG waveform.0.94ICC, higher is better#1 of 1
MESA macro-F1Sleep apnea (OSA) screeningOxiNetCaveat: Severity from estimated AHI.0.75macro-F1, higher is better#1 of 3
MrOS ICCSleep apnea (OSA) screeningOxiNetCaveat: External.0.94ICC, higher is better#1 of 1
MrOS macro-F1Sleep apnea (OSA) screeningOxiNetCaveat: External.0.80macro-F1, higher is better#1 of 1
SHHS1 ICCSleep apnea (OSA) screeningOxiNetCaveat: Same cohort as training; SHHS2 0.95.0.96ICC, higher is better#1 of 1
SHHS1 macro-F1Sleep apnea (OSA) screeningOxiNetCaveat: Same cohort as training; SHHS2 0.83.0.84macro-F1, higher is better#1 of 1

8 rows are credited to this paper. Ranks apply within each leaderboard only, as reported by the source.

Tasks & signals

Tasks

Sleep apnea (OSA) screeningSleep apneaSpO₂

Signals

SpO₂

Keywords

Sleep ApneaOSAAHIOximetrySpO₂NSRRExternal ValidationFairness

Datasets used

How it works

How Method works: Setup → Method → TargetSetup: Signals (SpO2); Form factor (finger). Method: Method. Target: Sleep apnea; SpO₂.SetupSignals: SpO2R / IRSignalsSpO2Form factor: fingerForm factorfingerMethodMethodTargetSleep apneaSleep apneaSpO₂O₂SpO₂

Cite this paper

Cite this paper with the fields recorded here. Check it against the publisher before submitting.

Levy, J., Álvarez, D., Del Campo, F., & Behar, J. A. (2023). Deep learning for obstructive sleep apnea diagnosis based on single channel oximetry. Nature Communications 14, 2023. https://doi.org/10.1038/s41467-023-40604-3
Levy, Jeremy, Daniel Álvarez, Félix Del Campo, and Joachim A. Behar. 2023. "Deep learning for obstructive sleep apnea diagnosis based on single channel oximetry." Nature Communications 14, 2023. https://doi.org/10.1038/s41467-023-40604-3.
Levy, J., Álvarez, D., Del Campo, F. and Behar, J. A. (2023) 'Deep learning for obstructive sleep apnea diagnosis based on single channel oximetry', Nature Communications 14, 2023. doi: 10.1038/s41467-023-40604-3.
Levy, Jeremy, et al. "Deep learning for obstructive sleep apnea diagnosis based on single channel oximetry." Nature Communications 14, 2023, 2023. https://doi.org/10.1038/s41467-023-40604-3.
J. Levy, D. Álvarez, F. Del Campo, and J. A. Behar, "Deep learning for obstructive sleep apnea diagnosis based on single channel oximetry," Nature Communications 14, 2023, 2023, doi: 10.1038/s41467-023-40604-3.
@article{levy2023deep,
  title         = {{Deep learning for obstructive sleep apnea diagnosis based on single channel oximetry}},
  author        = {Levy, Jeremy and Álvarez, Daniel and Del Campo, Félix and Behar, Joachim A.},
  year          = {2023},
  journal       = {Nature Communications 14, 2023},
  doi           = {10.1038/s41467-023-40604-3}
}

Added to the ChatPPG research hub

Same tasks, data and authors

Papers

Models

Datasets