Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Transfer learning for predicting acute myocardial infarction using electrocardiograms

Nyström, Axel LU ; Björkelund, Anders LU ; Ohlsson, Mattias LU orcid ; Björk, Jonas LU orcid ; Ekelund, Ulf LU orcid and Lundager Forberg, Jakob LU (2025) In PLOS Digital Health 4(10). p.0001058-0001058
Abstract

At the emergency department, it is important to quickly and accurately identify patients at risk of acute myocardial infarction (AMI). One of the main tools for detecting AMI is the electrocardiogram (ECG), which can be difficult to interpret manually. There is a long history of applying machine learning algorithms to ECGs, but such algorithms are quite data hungry, and correctly labeled high-quality ECGs are difficult to obtain. Transfer learning has been a successful strategy for mitigating data requirements in other applications, but the benefits for predicting AMI are understudied. Here we show that a straightforward application of transfer learning leads to large improvements also in this domain. We pre-train models to classify sex... (More)

At the emergency department, it is important to quickly and accurately identify patients at risk of acute myocardial infarction (AMI). One of the main tools for detecting AMI is the electrocardiogram (ECG), which can be difficult to interpret manually. There is a long history of applying machine learning algorithms to ECGs, but such algorithms are quite data hungry, and correctly labeled high-quality ECGs are difficult to obtain. Transfer learning has been a successful strategy for mitigating data requirements in other applications, but the benefits for predicting AMI are understudied. Here we show that a straightforward application of transfer learning leads to large improvements also in this domain. We pre-train models to classify sex and age using a collection of 840 k ECGs from non-chest-pain patients, and fine-tune the resulting models to predict AMI using 44 k ECGs from chest-pain patients. The results are compared with models trained without transfer learning. We find a considerable improvement from transfer learning, consistent across multiple state-of-the-art ResNet architectures and data sizes, with the best performing model improving from 0.79 AUC to 0.85 AUC. This suggests that even a simple form of transfer learning from a moderately sized dataset of non-chest-pain ECGs can lead to major improvements in predicting AMI.

(Less)
Please use this url to cite or link to this publication:
author
; ; ; ; and
organization
publishing date
type
Contribution to journal
publication status
published
subject
in
PLOS Digital Health
volume
4
issue
10
pages
0001058 - 0001058
publisher
Public Library of Science
external identifiers
  • scopus:105020647027
  • pmid:41171874
DOI
10.1371/journal.pdig.0001058
project
AIR Lund - Artificially Intelligent use of Registers
language
English
LU publication?
yes
additional info
Copyright: © 2025 Nyström et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
id
05c63886-1321-473e-a425-cd9ea64cc80c
date added to LUP
2025-12-01 11:20:46
date last changed
2025-12-16 05:33:37
@article{05c63886-1321-473e-a425-cd9ea64cc80c,
  abstract     = {{<p>At the emergency department, it is important to quickly and accurately identify patients at risk of acute myocardial infarction (AMI). One of the main tools for detecting AMI is the electrocardiogram (ECG), which can be difficult to interpret manually. There is a long history of applying machine learning algorithms to ECGs, but such algorithms are quite data hungry, and correctly labeled high-quality ECGs are difficult to obtain. Transfer learning has been a successful strategy for mitigating data requirements in other applications, but the benefits for predicting AMI are understudied. Here we show that a straightforward application of transfer learning leads to large improvements also in this domain. We pre-train models to classify sex and age using a collection of 840 k ECGs from non-chest-pain patients, and fine-tune the resulting models to predict AMI using 44 k ECGs from chest-pain patients. The results are compared with models trained without transfer learning. We find a considerable improvement from transfer learning, consistent across multiple state-of-the-art ResNet architectures and data sizes, with the best performing model improving from 0.79 AUC to 0.85 AUC. This suggests that even a simple form of transfer learning from a moderately sized dataset of non-chest-pain ECGs can lead to major improvements in predicting AMI.</p>}},
  author       = {{Nyström, Axel and Björkelund, Anders and Ohlsson, Mattias and Björk, Jonas and Ekelund, Ulf and Lundager Forberg, Jakob}},
  language     = {{eng}},
  number       = {{10}},
  pages        = {{0001058--0001058}},
  publisher    = {{Public Library of Science}},
  series       = {{PLOS Digital Health}},
  title        = {{Transfer learning for predicting acute myocardial infarction using electrocardiograms}},
  url          = {{http://dx.doi.org/10.1371/journal.pdig.0001058}},
  doi          = {{10.1371/journal.pdig.0001058}},
  volume       = {{4}},
  year         = {{2025}},
}