Cargando…
Patient contrastive learning: A performant, expressive, and practical approach to electrocardiogram modeling
Supervised machine learning applications in health care are often limited due to a scarcity of labeled training data. To mitigate the effect of small sample size, we introduce a pre-training approach, Patient Contrastive Learning of Representations (PCLR), which creates latent representations of ele...
Autores principales: | , , , , , |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
Public Library of Science
2022
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8880931/ https://www.ncbi.nlm.nih.gov/pubmed/35157695 http://dx.doi.org/10.1371/journal.pcbi.1009862 |
_version_ | 1784659353254494208 |
---|---|
author | Diamant, Nathaniel Reinertsen, Erik Song, Steven Aguirre, Aaron D. Stultz, Collin M. Batra, Puneet |
author_facet | Diamant, Nathaniel Reinertsen, Erik Song, Steven Aguirre, Aaron D. Stultz, Collin M. Batra, Puneet |
author_sort | Diamant, Nathaniel |
collection | PubMed |
description | Supervised machine learning applications in health care are often limited due to a scarcity of labeled training data. To mitigate the effect of small sample size, we introduce a pre-training approach, Patient Contrastive Learning of Representations (PCLR), which creates latent representations of electrocardiograms (ECGs) from a large number of unlabeled examples using contrastive learning. The resulting representations are expressive, performant, and practical across a wide spectrum of clinical tasks. We develop PCLR using a large health care system with over 3.2 million 12-lead ECGs and demonstrate that training linear models on PCLR representations achieves a 51% performance increase, on average, over six training set sizes and four tasks (sex classification, age regression, and the detection of left ventricular hypertrophy and atrial fibrillation), relative to training neural network models from scratch. We also compared PCLR to three other ECG pre-training approaches (supervised pre-training, unsupervised pre-training with an autoencoder, and pre-training using a contrastive multi ECG-segment approach), and show significant performance benefits in three out of four tasks. We found an average performance benefit of 47% over the other models and an average of a 9% performance benefit compared to best model for each task. We release PCLR to enable others to extract ECG representations at https://github.com/broadinstitute/ml4h/tree/master/model_zoo/PCLR. |
format | Online Article Text |
id | pubmed-8880931 |
institution | National Center for Biotechnology Information |
language | English |
publishDate | 2022 |
publisher | Public Library of Science |
record_format | MEDLINE/PubMed |
spelling | pubmed-88809312022-02-26 Patient contrastive learning: A performant, expressive, and practical approach to electrocardiogram modeling Diamant, Nathaniel Reinertsen, Erik Song, Steven Aguirre, Aaron D. Stultz, Collin M. Batra, Puneet PLoS Comput Biol Research Article Supervised machine learning applications in health care are often limited due to a scarcity of labeled training data. To mitigate the effect of small sample size, we introduce a pre-training approach, Patient Contrastive Learning of Representations (PCLR), which creates latent representations of electrocardiograms (ECGs) from a large number of unlabeled examples using contrastive learning. The resulting representations are expressive, performant, and practical across a wide spectrum of clinical tasks. We develop PCLR using a large health care system with over 3.2 million 12-lead ECGs and demonstrate that training linear models on PCLR representations achieves a 51% performance increase, on average, over six training set sizes and four tasks (sex classification, age regression, and the detection of left ventricular hypertrophy and atrial fibrillation), relative to training neural network models from scratch. We also compared PCLR to three other ECG pre-training approaches (supervised pre-training, unsupervised pre-training with an autoencoder, and pre-training using a contrastive multi ECG-segment approach), and show significant performance benefits in three out of four tasks. We found an average performance benefit of 47% over the other models and an average of a 9% performance benefit compared to best model for each task. We release PCLR to enable others to extract ECG representations at https://github.com/broadinstitute/ml4h/tree/master/model_zoo/PCLR. Public Library of Science 2022-02-14 /pmc/articles/PMC8880931/ /pubmed/35157695 http://dx.doi.org/10.1371/journal.pcbi.1009862 Text en © 2022 Diamant et al https://creativecommons.org/licenses/by/4.0/This is an open access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/) , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. |
spellingShingle | Research Article Diamant, Nathaniel Reinertsen, Erik Song, Steven Aguirre, Aaron D. Stultz, Collin M. Batra, Puneet Patient contrastive learning: A performant, expressive, and practical approach to electrocardiogram modeling |
title | Patient contrastive learning: A performant, expressive, and practical approach to electrocardiogram modeling |
title_full | Patient contrastive learning: A performant, expressive, and practical approach to electrocardiogram modeling |
title_fullStr | Patient contrastive learning: A performant, expressive, and practical approach to electrocardiogram modeling |
title_full_unstemmed | Patient contrastive learning: A performant, expressive, and practical approach to electrocardiogram modeling |
title_short | Patient contrastive learning: A performant, expressive, and practical approach to electrocardiogram modeling |
title_sort | patient contrastive learning: a performant, expressive, and practical approach to electrocardiogram modeling |
topic | Research Article |
url | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8880931/ https://www.ncbi.nlm.nih.gov/pubmed/35157695 http://dx.doi.org/10.1371/journal.pcbi.1009862 |
work_keys_str_mv | AT diamantnathaniel patientcontrastivelearningaperformantexpressiveandpracticalapproachtoelectrocardiogrammodeling AT reinertsenerik patientcontrastivelearningaperformantexpressiveandpracticalapproachtoelectrocardiogrammodeling AT songsteven patientcontrastivelearningaperformantexpressiveandpracticalapproachtoelectrocardiogrammodeling AT aguirreaarond patientcontrastivelearningaperformantexpressiveandpracticalapproachtoelectrocardiogrammodeling AT stultzcollinm patientcontrastivelearningaperformantexpressiveandpracticalapproachtoelectrocardiogrammodeling AT batrapuneet patientcontrastivelearningaperformantexpressiveandpracticalapproachtoelectrocardiogrammodeling |