Academic Bibliography

 Search 200 years of publications by Ghent University researchers.
Advanced search
2 files | 6.28 MB
Add to list
  1. Search results
  2. Do not sleep on traditional machine l...

Do not sleep on traditional machine learning simple and interpretable techniques are competitive to deep learning for sleep scoring

Jeroen Van Der Donckt (UGent) , Jonas Van Der Donckt (UGent) , Emiel Deprost (UGent) , Nicolas Vandenbussche (UGent) , Michael Rademaker, Gilles Vandewiele (UGent) and Sofie Van Hoecke (UGent)
(2023) BIOMEDICAL SIGNAL PROCESSING AND CONTROL. 81.
Author
Organization
Project
Abstract
Over the last few years, research in automatic sleep scoring has mainly focused on developing increasingly complex deep learning architectures. However, recently these approaches achieved only marginal improve-ments, often at the expense of requiring more data and more expensive training procedures. Despite all these efforts and their satisfactory performance, automatic sleep staging solutions are not widely adopted in a clinical context yet. We argue that most deep learning solutions for sleep scoring are limited in their real-world applicability as they are hard to train, deploy, and reproduce. Moreover, these solutions lack interpretability and transparency, which are often key to increase adoption rates. In this work, we revisit the problem of sleep stage classification using classical machine learning. Results show that competitive performance can be achieved with a conventional machine learning pipeline consisting of preprocessing, feature extraction, and a simple machine learning model. In particular, we analyze the performance of a linear model and a non-linear (gradient boosting) model. Our approach surpasses state-of-the-art (that uses the same data) on two public datasets: Sleep-EDF SC-20 (MF1 0.810) and Sleep-EDF ST (MF1 0.795), while achieving competitive results on Sleep-EDF SC-78 (MF1 0.775) and MASS SS3 (MF1 0.817). We show that, for the sleep stage scoring task, the expressiveness of an engineered feature vector is on par with the internally learned representations of deep learning models. This observation opens the door to clinical adoption, as a representative feature vector allows to leverage both the interpretability and successful track record of traditional machine learning models.
Keywords
Sleep scoring, Time series, Machine learning, Open source, STAGE CLASSIFICATION, NEURAL-NETWORK, SYSTEM, HEALTH

Downloads

  • Download
    Sleep Linear Model revision 2 1 file.pdf
    • full text (Accepted manuscript)
    • |
    • open access
    • |
    • PDF
    • |
    • 4.89 MB
  • Download
    (...).pdf
    • full text (Published version)
    • |
    • UGent only
    • |
    • PDF
    • |
    • 1.39 MB

Citation

Please use this url to cite or link to this publication:

MLA
Van Der Donckt, Jeroen, et al. “Do Not Sleep on Traditional Machine Learning Simple and Interpretable Techniques Are Competitive to Deep Learning for Sleep Scoring.” BIOMEDICAL SIGNAL PROCESSING AND CONTROL, vol. 81, 2023, doi:10.1016/j.bspc.2022.104429.
APA
Van Der Donckt, J., Van Der Donckt, J., Deprost, E., Vandenbussche, N., Rademaker, M., Vandewiele, G., & Van Hoecke, S. (2023). Do not sleep on traditional machine learning simple and interpretable techniques are competitive to deep learning for sleep scoring. BIOMEDICAL SIGNAL PROCESSING AND CONTROL, 81. https://doi.org/10.1016/j.bspc.2022.104429
Chicago author-date
Van Der Donckt, Jeroen, Jonas Van Der Donckt, Emiel Deprost, Nicolas Vandenbussche, Michael Rademaker, Gilles Vandewiele, and Sofie Van Hoecke. 2023. “Do Not Sleep on Traditional Machine Learning Simple and Interpretable Techniques Are Competitive to Deep Learning for Sleep Scoring.” BIOMEDICAL SIGNAL PROCESSING AND CONTROL 81. https://doi.org/10.1016/j.bspc.2022.104429.
Chicago author-date (all authors)
Van Der Donckt, Jeroen, Jonas Van Der Donckt, Emiel Deprost, Nicolas Vandenbussche, Michael Rademaker, Gilles Vandewiele, and Sofie Van Hoecke. 2023. “Do Not Sleep on Traditional Machine Learning Simple and Interpretable Techniques Are Competitive to Deep Learning for Sleep Scoring.” BIOMEDICAL SIGNAL PROCESSING AND CONTROL 81. doi:10.1016/j.bspc.2022.104429.
Vancouver
1.
Van Der Donckt J, Van Der Donckt J, Deprost E, Vandenbussche N, Rademaker M, Vandewiele G, et al. Do not sleep on traditional machine learning simple and interpretable techniques are competitive to deep learning for sleep scoring. BIOMEDICAL SIGNAL PROCESSING AND CONTROL. 2023;81.
IEEE
[1]
J. Van Der Donckt et al., “Do not sleep on traditional machine learning simple and interpretable techniques are competitive to deep learning for sleep scoring,” BIOMEDICAL SIGNAL PROCESSING AND CONTROL, vol. 81, 2023.
@article{01GPJN2K19T32MTA1THNVN91JZ,
  abstract     = {{Over the last few years, research in automatic sleep scoring has mainly focused on developing increasingly complex deep learning architectures. However, recently these approaches achieved only marginal improve-ments, often at the expense of requiring more data and more expensive training procedures. Despite all these efforts and their satisfactory performance, automatic sleep staging solutions are not widely adopted in a clinical context yet. We argue that most deep learning solutions for sleep scoring are limited in their real-world applicability as they are hard to train, deploy, and reproduce. Moreover, these solutions lack interpretability and transparency, which are often key to increase adoption rates. In this work, we revisit the problem of sleep stage classification using classical machine learning. Results show that competitive performance can be achieved with a conventional machine learning pipeline consisting of preprocessing, feature extraction, and a simple machine learning model. In particular, we analyze the performance of a linear model and a non-linear (gradient boosting) model. Our approach surpasses state-of-the-art (that uses the same data) on two public datasets: Sleep-EDF SC-20 (MF1 0.810) and Sleep-EDF ST (MF1 0.795), while achieving competitive results on Sleep-EDF SC-78 (MF1 0.775) and MASS SS3 (MF1 0.817). We show that, for the sleep stage scoring task, the expressiveness of an engineered feature vector is on par with the internally learned representations of deep learning models. This observation opens the door to clinical adoption, as a representative feature vector allows to leverage both the interpretability and successful track record of traditional machine learning models.}},
  articleno    = {{104429}},
  author       = {{Van Der Donckt, Jeroen and Van Der Donckt, Jonas and Deprost, Emiel and Vandenbussche, Nicolas and  Rademaker, Michael and Vandewiele, Gilles and Van Hoecke, Sofie}},
  issn         = {{1746-8094}},
  journal      = {{BIOMEDICAL SIGNAL PROCESSING AND CONTROL}},
  keywords     = {{Sleep scoring,Time series,Machine learning,Open source,STAGE CLASSIFICATION,NEURAL-NETWORK,SYSTEM,HEALTH}},
  language     = {{eng}},
  pages        = {{9}},
  title        = {{Do not sleep on traditional machine learning simple and interpretable techniques are competitive to deep learning for sleep scoring}},
  url          = {{http://doi.org/10.1016/j.bspc.2022.104429}},
  volume       = {{81}},
  year         = {{2023}},
}
Altmetric
View in Altmetric
Web of Science
Times cited: