EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Compressed Deep Learning Models for Wearable Atrial Fibrillation Detection through Attention

Marko Mäkynen, G Andre Ng iD, Xin Li, Fernando S Schlindwein iD, Timothy C Pearce iD

Sensors · 2024

DOI: 10.3390/s24154787 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, 2024. Verfasst von Marko Mäkynen und 4 weiteren Autor:innen. Beteiligte Einrichtung: Biomedical Engineering Research Group, School of Engineering, University of Leicester, Leicester LE1 7RH, UK. Hauptthemen laut MEDLINE: Atrial Fibrillation, Deep Learning, Wearable Electronic Devices, Electrocardiography. Die Arbeit wurde bisher 13-mal zitiert (Crossref) und nennt 52 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: NIHR Leicester Biomedical Research Centre, British Heart Foundation, Medical Research Council Biomedical Catalyst Developmental Pathway Funding Scheme.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Deep learning (DL) models have shown promise for the accurate detection of atrial fibrillation (AF) from electrocardiogram/photoplethysmography (ECG/PPG) data, yet deploying these on resource-constrained wearable devices remains challenging. This study proposes integrating a customized channel attention mechanism to compress DL neural networks for AF detection, allowing the model to focus only on the most salient time-series features. The results demonstrate that applying compression through channel attention significantly reduces the total number of model parameters and file size while minimizing loss in detection accuracy. Notably, after compression, performance increases for certain model variants in key AF databases (ADB and C2017DB). Moreover, analyzing the learned channel attention distributions after training enhances the explainability of the AF detection models by highlighting the salient temporal ECG/PPG features most important for its diagnosis. Overall, this research establishes that integrating attention mechanisms is an effective strategy for compressing large DL models, making them deployable on low-power wearable devices. We show that this approach yields compressed, accurate, and explainable AF detectors ideal for wearables. Incorporating channel attention enables simpler yet more accurate algorithms that have the potential to provide clinicians with valuable insights into the salient temporal biomarkers of AF. Our findings highlight that the use of attention is an important direction for the future development of efficient, high-performing, and interpretable AF screening tools for wearable technology.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Atrial FibrillationDeep LearningWearable Electronic DevicesElectrocardiographyAlgorithmsNeural Networks, ComputerHumansPhotoplethysmographySignal Processing, Computer-Assisted

Stichwörter der Autor:innen

artificial intelligenceatrial fibrillationdeep learningwearable devices

Autor:innen und Institutionen

  • Biomedical Engineering Research Group, School of Engineering, University of Leicester, Leicester LE1 7RH, UK.
  • National Institute for Health Research Leicester Cardiovascular Biomedical Research Centre, Glenfield Hospital, Leicester LE5 4PW, UK.
  • Department of Cardiovascular Sciences, University of Leicester, Leicester LE1 7RH, UK.

ORCID-Profile: G Andre Ng, Fernando S Schlindwein, Timothy C Pearce

Förderung

  • NIHR Leicester Biomedical Research Centre
  • British Heart Foundation
  • Medical Research Council Biomedical Catalyst Developmental Pathway Funding Scheme

Zitieren

APA

Mäkynen, M., Ng, G. A., Li, X., Schlindwein, F. S., & Pearce, T. C. (2024). Compressed Deep Learning Models for Wearable Atrial Fibrillation Detection through Attention. Sensors. https://doi.org/10.3390/s24154787

Vancouver

Mäkynen M, Ng GA, Li X, Schlindwein FS, Pearce TC. Compressed Deep Learning Models for Wearable Atrial Fibrillation Detection through Attention. Sensors. 2024. doi:10.3390/s24154787

BibTeX
@article{Mkynen2024,
  author = {Mäkynen, Marko and Ng, G Andre and Li, Xin and Schlindwein, Fernando S and Pearce, Timothy C},
  title = {Compressed Deep Learning Models for Wearable Atrial Fibrillation Detection through Attention},
  journal = {Sensors},
  year = {2024},
  doi = {10.3390/s24154787}
}
Kostenlos starten →