EUVIMEDEUVIMED entdecken
Crossref

An Interpretable Multiwavelet Neural Network for Automatic Sleep Staging

Kairui Wu, Binyang Ding, Hanfei Zhu, Yuhao Wu, Yongming Han, Tingrui Pan

IEEE Sensors Journal · 2026-08 · 26 (15) · 22887-22898

DOI: 10.1109/jsen.2026.3696451 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Sensors Journal, 2026-08. Verfasst von Kairui Wu und 5 weiteren Autor:innen. Die Arbeit nennt 50 Literaturangaben. Gefördert von: Key Technologies Research and Development Program of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Deep learning (DL) has been widely applied to automatic sleep staging due to its strong capability for feature representation. However, most existing DL-based approaches do not explicitly incorporate medical knowledge or physiological mechanisms into the learning process, which raises concerns regarding their reliability and robustness in clinical applications. To address this limitation, this article proposes a novel interpretable DL model for sleep staging, termed the Multi-Core Wavelet Transform Network (MCTNet). To detect and localize sleep-related characteristic waves, a wavelet feature module (WFM) is constructed based on the continuous wavelet transform (CWT). Furthermore, to coordinate the contributions of different wavelet kernels and emphasize discriminative sleep features, a power spectral density (PSD)-based weight calibration module (PWCM) is designed to dynamically assign adaptive weights to different wavelet kernels. The effectiveness of MCTNet is demonstrated by its state-of-the-art performance on two publicly available datasets: 1) SleepEDF-78 and 2) SleepEDF-20. Wavelet kernel ablation experiments further validate the specificity and effectiveness of the proposed wavelet kernels. In addition, the interpretability of the proposed method is illustrated through feature heatmap visualizations and variations in kernel weights, indicating that MCTNet provides physiologically meaningful feature representations that can offer valuable references for clinical decision-making.

Förderung

  • Key Technologies Research and Development Program of China

Zitieren

APA

Wu, K., Ding, B., Zhu, H., Wu, Y., Han, Y., & Pan, T. (2026). An Interpretable Multiwavelet Neural Network for Automatic Sleep Staging. IEEE Sensors Journal, 26(15), 22887-22898. https://doi.org/10.1109/jsen.2026.3696451

Vancouver

Wu K, Ding B, Zhu H, Wu Y, Han Y, Pan T. An Interpretable Multiwavelet Neural Network for Automatic Sleep Staging. IEEE Sensors Journal. 2026;26(15):22887-22898. doi:10.1109/jsen.2026.3696451

BibTeX
@article{Wu2026,
  author = {Wu, Kairui and Ding, Binyang and Zhu, Hanfei and Wu, Yuhao and Han, Yongming and Pan, Tingrui},
  title = {An Interpretable Multiwavelet Neural Network for Automatic Sleep Staging},
  journal = {IEEE Sensors Journal},
  year = {2026},
  volume = {26},
  number = {15},
  pages = {22887--22898},
  doi = {10.1109/jsen.2026.3696451}
}
Kostenlos starten →