Paperdoi:10.1093/sleep/zsaa098

Deep learning enables sleep staging from photoplethysmogram for patients with suspected sleep apnea

Venue listed2020Added Sep 26, 2026

Authors: Henri Korkalainen, Juhani Aakko, Brett Duce, Samu Kainulainen, Akseli Leino, Sami Nikkonen, Isaac O. Afara, Sami Myllymaa et al. (10 authors)

Sleep

Publisher Cite

Henri Korkalainen et al., "Deep learning enables sleep staging from photoplethysmogram for patients with suspected sleep apnea", doi:10.1093/sleep/zsaa098, CC BY-NC 4.0

TL;DR

CNN+RNN sleep staging from finger pulse-oximeter PPG in n=894 suspected OSA patients from diagnostic PSG. Models for 3-class (wake/NREM/REM), 4-class (wake/N1+N2/N3/REM), and 5-class (wake/N1/N2/N3/REM). Landmark clinical demonstration that PPG alone can estimate sleep time/efficiency and differentiate stages with moderate agreement to manual EEG scoring in an OSA-suspected population.

Form factor: finger

Abstract

Study Objectives
Accurate identification of sleep stages is essential in the diagnosis of sleep disorders (e.g. obstructive sleep apnea [OSA]) but relies on labor-intensive electroencephalogram (EEG)-based manual scoring. Furthermore, long-term assessment of sleep relies on actigraphy differentiating only between wake and sleep periods without identifying specific sleep stages and having low reliability in identifying wake periods after sleep onset. To address these issues, we aimed to develop an automatic method for identifying the sleep stages from the photoplethysmogram (PPG) signal obtained with a simple finger pulse oximeter.
Methods
PPG signals from the diagnostic polysomnographies of susptected OSA patients (n = 894) were utilized to develop a combined convolutional and recurrent neural network. The deep learning model was trained individually for three-stage (wake/NREM/REM), four-stage (wake/N1+N2/N3/REM), and five-stage (wake/N1/N2/N3/REM) classification of sleep.
Results
The three-stage model achieved an epoch-by-epoch accuracy of 80.1% with Cohen’s κ of 0.65. The four- and five-stage models achieved 68.5% (κ = 0.54), and 64.1% (κ = 0.51) accuracies, respectively. With the five-stage model, the total sleep time was underestimated with a mean bias error (SD) of of 7.5 (55.2) minutes.
Conclusion
The PPG-based deep learning model enabled accurate estimation of sleep time and differentiation between sleep stages with a moderate agreement to manual EEG-based scoring. As PPG is already included in ambulatory polygraphic recordings, applying the PPG-based sleep staging could improve their diagnostic value by enabling simple, low-cost, and reliable monitoring of sleep and help assess otherwise overlooked conditions such as REM-related OSA.

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

Results on ChatPPG leaderboards

All leaderboards

No results from this paper on ChatPPG leaderboards yet. Browse the leaderboards.

Tasks & signals

Tasks

Sleep staging

Signals

PPG

Keywords

sleepstagingOSAfingerlandmark

How it works

Wavelengths: red, IR

Cite this paper

APA

Korkalainen, H., Aakko, J., Duce, B., Kainulainen, S., Leino, A., Nikkonen, S., Afara, I. O., Myllymaa, S., et al. (2020). Deep learning enables sleep staging from photoplethysmogram for patients with suspected sleep apnea. Sleep. https://doi.org/10.1093/sleep/zsaa098

BibTeX

@article{korkalainen2020deep,
  title         = {{Deep learning enables sleep staging from photoplethysmogram for patients with suspected sleep apnea}},
  author        = {Korkalainen, Henri and Aakko, Juhani and Duce, Brett and Kainulainen, Samu and Leino, Akseli and Nikkonen, Sami and Afara, Isaac O. and Myllymaa, Sami and others},
  year          = {2020},
  journal       = {Sleep},
  doi           = {10.1093/sleep/zsaa098}
}

Built from the hub record. Check it against the publisher before submitting.

Sep 26, 2026Added to the ChatPPG research hub