EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Usefulness of Heat Map Explanations for Deep-Learning-Based Electrocardiogram Analysis

Andrea M Storås iD, Ole Emil Andersen, Sam Lockhart, Roman Thielemann iD, Filip Gnesin, Vajira Thambawita iD, Steven A Hicks, Jørgen K Kanters iD, Inga Strümke iD, Pål Halvorsen iD, Michael A Riegler

Diagnostics · 11.07.2023 · 13 (14) · 2345

DOI: 10.3390/diagnostics13142345 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Diagnostics, online erschienen am 11.07.2023. Verfasst von Andrea M Storås und 10 weiteren Autor:innen. Beteiligte Einrichtung: Department of Holistic Systems, Simula Metropolitan Center for Digital Engineering, 0167 Oslo, Norway. Die Arbeit wurde bisher 13-mal zitiert (Crossref) und nennt 31 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Wellcome Trust Clinical PhD Fellowship.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Deep neural networks are complex machine learning models that have shown promising results in analyzing high-dimensional data such as those collected from medical examinations. Such models have the potential to provide fast and accurate medical diagnoses. However, the high complexity makes deep neural networks and their predictions difficult to understand. Providing model explanations can be a way of increasing the understanding of “black box” models and building trust. In this work, we applied transfer learning to develop a deep neural network to predict sex from electrocardiograms. Using the visual explanation method Grad-CAM, heat maps were generated from the model in order to understand how it makes predictions. To evaluate the usefulness of the heat maps and determine if the heat maps identified electrocardiogram features that could be recognized to discriminate sex, medical doctors provided feedback. Based on the feedback, we concluded that, in our setting, this mode of explainable artificial intelligence does not provide meaningful information to medical doctors and is not useful in the clinic. Our results indicate that improved explanation techniques that are tailored to medical data should be developed before deep neural networks can be applied in the clinic for diagnostic purposes.

Schlagwörter

Stichwörter der Autor:innen

electrocardiogramsexplainable artificial intelligenceheat maps

Autor:innen und Institutionen

  • Department of Holistic Systems, Simula Metropolitan Center for Digital Engineering, 0167 Oslo, Norway.
  • Department of Computer Science, Oslo Metropolitan University, 0130 Oslo, Norway.
  • Department of Public Health, Aarhus University, 8000 Aarhus, Denmark.
  • Steno Diabetes Center, Aarhus University, 8000 Aarhus, Denmark.
  • Wellcome Trust-Medical Research Council Institute of Metabolic Science, University of Cambridge, Cambridge CB2 0QQ, UK.
  • Novo Nordisk Foundation Center for Basic Metabolic Research, University of Copenhagen, 2200 Copenhagen, Denmark.

ORCID-Profile: Andrea M Storås, Roman Thielemann, Vajira Thambawita, Jørgen K Kanters, Inga Strümke, Pål Halvorsen

Förderung

  • Wellcome Trust Clinical PhD Fellowship

Zitieren

APA

Storås, A. M., Andersen, O. E., Lockhart, S., Thielemann, R., Gnesin, F., Thambawita, V., Hicks, S. A., Kanters, J. K., Strümke, I., Halvorsen, P., & Riegler, M. A. (2023). Usefulness of Heat Map Explanations for Deep-Learning-Based Electrocardiogram Analysis. Diagnostics, 13(14), 2345. https://doi.org/10.3390/diagnostics13142345

Vancouver

Storås AM, Andersen OE, Lockhart S, Thielemann R, Gnesin F, Thambawita V, et al. Usefulness of Heat Map Explanations for Deep-Learning-Based Electrocardiogram Analysis. Diagnostics. 2023;13(14):2345. doi:10.3390/diagnostics13142345

BibTeX
@article{Stors2023,
  author = {Storås, Andrea M and Andersen, Ole Emil and Lockhart, Sam and Thielemann, Roman and Gnesin, Filip and Thambawita, Vajira and Hicks, Steven A and Kanters, Jørgen K and Strümke, Inga and Halvorsen, Pål and Riegler, Michael A},
  title = {Usefulness of Heat Map Explanations for Deep-Learning-Based Electrocardiogram Analysis},
  journal = {Diagnostics},
  year = {2023},
  volume = {13},
  number = {14},
  pages = {2345},
  doi = {10.3390/diagnostics13142345}
}
Kostenlos starten →