EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

SleepStageNet: A Lightweight and Explainable Deep Learning Architecture for Multi-Channel Sleep Staging

Ali Alhazmi iD

Brain Sciences · 14.08.2026 · 16 (8) · 860

DOI: 10.3390/brainsci16080860 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Brain Sciences, online erschienen am 14.08.2026. Verfasst von Ali Alhazmi. Beteiligte Einrichtung: Department of Computer Science, College of Engineering and Computer Science, Jazan University, Jazan 45142, Saudi Arabia. Die Arbeit nennt 35 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Jazan University.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background/Objectives: Automatic sleep staging from polysomnography (PSG) is particularly challenging in clinical cohorts with neurological disorders.

Methods

This study presents SleepStageNet, a compact model (1.075M parameters) that integrates a dual-branch convolutional epoch encoder, feature gating, a bidirectional gated recurrent unit, and multi-head self-attention for five-class staging from five PSG channels (C3, C4, EOG1, EOG2, and chin EMG). The individual operations are adapted from established architectures; the study contribution is their compact integration and controlled evaluation in an Indian acute stroke cohort.

Results

Of the 100 recordings in the Indian Sleep Polysomnography (iSLEEPS) resource, 95 satisfied the five-channel extraction criteria, yielding 78,323 annotated epochs. Subject-independent 10-fold stratified group cross-validation produced an accuracy of 73.91 ± 2.24%, a macro F1-score of 67.29 ± 1.96%, and a Cohen’s κ of 0.634±0.029 (sample standard deviations). A matched single-branch encoder obtained κ=0.635 (full minus single branch: Δκ=−0.001, Holm-adjusted p=0.846), while matched C4-only input obtained κ=0.600 (full minus C4-only: Δκ=0.033, Holm-adjusted p=0.008). Grad-CAM and temporal attention visualizations provided qualitative evidence of physiologically plausible focus, while channel occlusion quantified the contribution of each signal. Without fine-tuning, a 10-model ensemble obtained κ=0.614 on ISRUC-SLEEP Subgroup III (10 healthy subjects; 8889 epochs).

Conclusions

These results establish a reproducible reference for this clinical cohort while identifying the need for broader external and prospective validation.

Schlagwörter

Stichwörter der Autor:innen

ISRUC-SLEEPIndian sleep datasetconvolutional neural networkcross-dataset validationdeep learningexplainabilitygated recurrent unitpolysomnographyself-attentionsleep stage classificationstroke patients

Autor:innen und Institutionen

  • Department of Computer Science, College of Engineering and Computer Science, Jazan University, Jazan 45142, Saudi Arabia.
  • Engineering and Technology Research Center, Jazan University, Jazan 82817, Saudi Arabia.

ORCID-Profile: Ali Alhazmi

Förderung

  • Jazan University

Zitieren

APA

Alhazmi, A. (2026). SleepStageNet: A Lightweight and Explainable Deep Learning Architecture for Multi-Channel Sleep Staging. Brain Sciences, 16(8), 860. https://doi.org/10.3390/brainsci16080860

Vancouver

Alhazmi A. SleepStageNet: A Lightweight and Explainable Deep Learning Architecture for Multi-Channel Sleep Staging. Brain Sciences. 2026;16(8):860. doi:10.3390/brainsci16080860

BibTeX
@article{Alhazmi2026,
  author = {Alhazmi, Ali},
  title = {SleepStageNet: A Lightweight and Explainable Deep Learning Architecture for Multi-Channel Sleep Staging},
  journal = {Brain Sciences},
  year = {2026},
  volume = {16},
  number = {8},
  pages = {860},
  doi = {10.3390/brainsci16080860}
}
Kostenlos starten →