{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,20]],"date-time":"2026-03-20T21:42:19Z","timestamp":1774042939287,"version":"3.50.1"},"update-to":[{"DOI":"10.1371\/journal.pcbi.1009862","type":"new_version","label":"New version","source":"publisher","updated":{"date-parts":[[2022,2,25]],"date-time":"2022-02-25T00:00:00Z","timestamp":1645747200000}}],"reference-count":34,"publisher":"Public Library of Science (PLoS)","issue":"2","license":[{"start":{"date-parts":[[2022,2,14]],"date-time":"2022-02-14T00:00:00Z","timestamp":1644796800000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"ibm"},{"DOI":"10.13039\/100004326","name":"bayer ag","doi-asserted-by":"crossref","id":[{"id":"10.13039\/100004326","id-type":"DOI","asserted-by":"crossref"}]},{"name":"quanta computing"},{"name":"Controlled Risk Insurance Company\/Risk Management Foundation"}],"content-domain":{"domain":["www.ploscompbiol.org"],"crossmark-restriction":false},"short-container-title":["PLoS Comput Biol"],"abstract":"<jats:p>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,<jats:bold>P<\/jats:bold>atient<jats:bold>C<\/jats:bold>ontrastive<jats:bold>L<\/jats:bold>earning of<jats:bold>R<\/jats:bold>epresentations (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<jats:ext-link xmlns:xlink=\"http:\/\/www.w3.org\/1999\/xlink\" ext-link-type=\"uri\" xlink:href=\"https:\/\/github.com\/broadinstitute\/ml4h\/tree\/master\/model_zoo\/PCLR\" xlink:type=\"simple\">https:\/\/github.com\/broadinstitute\/ml4h\/tree\/master\/model_zoo\/PCLR<\/jats:ext-link>.<\/jats:p>","DOI":"10.1371\/journal.pcbi.1009862","type":"journal-article","created":{"date-parts":[[2022,2,14]],"date-time":"2022-02-14T19:02:56Z","timestamp":1644865376000},"page":"e1009862","update-policy":"https:\/\/doi.org\/10.1371\/journal.pcbi.corrections_policy","source":"Crossref","is-referenced-by-count":41,"title":["Patient contrastive learning: A performant, expressive, and practical approach to electrocardiogram modeling"],"prefix":"10.1371","volume":"18","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-1738-304X","authenticated-orcid":true,"given":"Nathaniel","family":"Diamant","sequence":"first","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7214-9611","authenticated-orcid":true,"given":"Erik","family":"Reinertsen","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5868-3877","authenticated-orcid":true,"given":"Steven","family":"Song","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5509-1646","authenticated-orcid":true,"given":"Aaron D.","family":"Aguirre","sequence":"additional","affiliation":[]},{"given":"Collin M.","family":"Stultz","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6822-0593","authenticated-orcid":true,"given":"Puneet","family":"Batra","sequence":"additional","affiliation":[]}],"member":"340","published-online":{"date-parts":[[2022,2,14]]},"reference":[{"issue":"18","key":"pcbi.1009862.ref001","doi-asserted-by":"crossref","first-page":"2158","DOI":"10.1161\/hc4301.098254","article-title":"Sudden Cardiac Death in the United States, 1989 to 1998","volume":"104","author":"ZJ Zheng","year":"2001","journal-title":"Circulation"},{"issue":"2","key":"pcbi.1009862.ref002","doi-asserted-by":"crossref","first-page":"415","DOI":"10.1080\/10543400903572829","article-title":"Practical Issues in Building Risk-Predicting Models for Complex Diseases","volume":"20","author":"J Kang","year":"2010","journal-title":"Journal of Biopharmaceutical Statistics"},{"issue":"19","key":"pcbi.1009862.ref003","first-page":"625","article-title":"Why Does Unsupervised Pre-training Help Deep Learning?","volume":"11","author":"D Erhan","year":"2010","journal-title":"Journal of Machine Learning Research"},{"key":"pcbi.1009862.ref004","unstructured":"Erhan D, Manzagol PA, Bengio Y, Bengio S, Vincent P. The Difficulty of Training Deep Architectures and the Effect of Unsupervised Pre-Training. In: Proceedings of the Twelth International Conference on Artificial Intelligence and Statistics. PMLR; 2009. p. 153\u2013160."},{"key":"pcbi.1009862.ref005","doi-asserted-by":"crossref","unstructured":"Azizi S, Mustafa B, Ryan F, Beaver Z, Freyberg J, Deaton J, et al. Big Self-Supervised Models Advance Medical Image Classification. arXiv:210105224 [cs, eess]. 2021;.","DOI":"10.1109\/ICCV48922.2021.00346"},{"key":"pcbi.1009862.ref006","unstructured":"Kiyasseh D, Zhu T, Clifton DA. CLOCS: Contrastive Learning of Cardiac Signals across Space, Time, and Patients. In: Meila M, Zhang T, editors. Proceedings of the 38th International Conference on Machine Learning. vol. 139 of Proceedings of Machine Learning Research. PMLR; 2021. p. 5606\u20135615."},{"key":"pcbi.1009862.ref007","unstructured":"Chen T, Kornblith S, Norouzi M, Hinton G. A Simple Framework for Contrastive Learning of Visual Representations. In: Proceedings of the 37th International Conference on Machine Learning. PMLR; 2020. p. 1597\u20131607."},{"issue":"1","key":"pcbi.1009862.ref008","doi-asserted-by":"crossref","first-page":"1760","DOI":"10.1038\/s41467-020-15432-4","article-title":"Automatic Diagnosis of the 12-Lead ECG Using a Deep Neural Network","volume":"11","author":"AH Ribeiro","year":"2020","journal-title":"Nature Communications"},{"key":"pcbi.1009862.ref009","doi-asserted-by":"crossref","first-page":"100423","DOI":"10.1016\/j.ijcha.2019.100423","article-title":"A Deep Neural Network for 12-Lead Electrocardiogram Interpretation Outperforms a Conventional Algorithm, and Its Physician Overread, in the Diagnosis of Atrial Fibrillation","volume":"25","author":"SW Smith","year":"2019","journal-title":"IJC Heart & Vasculature"},{"issue":"1","key":"pcbi.1009862.ref010","doi-asserted-by":"crossref","first-page":"65","DOI":"10.1038\/s41591-018-0268-3","article-title":"Cardiologist-Level Arrhythmia Detection and Classification in Ambulatory Electrocardiograms Using a Deep Neural Network","volume":"25","author":"AY Hannun","year":"2019","journal-title":"Nature Medicine"},{"issue":"10","key":"pcbi.1009862.ref011","doi-asserted-by":"crossref","DOI":"10.1161\/JAHA.119.015138","article-title":"Automatic Triage of 12-Lead ECGs Using Deep Convolutional Neural Networks","volume":"9","author":"RR van de Leur","year":"2020","journal-title":"Journal of the American Heart Association"},{"issue":"9","key":"pcbi.1009862.ref012","article-title":"Age and Sex Estimation Using Artificial Intelligence From Standard 12-Lead ECGs","volume":"12","author":"ZI Attia","year":"2019","journal-title":"Circulation: Arrhythmia and Electrophysiology"},{"issue":"13","key":"pcbi.1009862.ref013","doi-asserted-by":"crossref","first-page":"1287","DOI":"10.1161\/CIRCULATIONAHA.120.047829","article-title":"Deep Neural Networks Can Predict New-Onset Atrial Fibrillation From the 12-Lead ECG and Help Identify Those at Risk of Atrial Fibrillation extendashRelated Stroke","volume":"143","author":"S Raghunath","year":"2021","journal-title":"Circulation"},{"key":"pcbi.1009862.ref014","doi-asserted-by":"crossref","first-page":"S104","DOI":"10.1016\/j.jelectrocard.2019.08.033","article-title":"Deep Neural Networks Can Predict One-Year Mortality and Incident Atrial Fibrillation from Raw 12-Lead Electrocardiogram Voltage Data","volume":"57","author":"S Raghunath","year":"2019","journal-title":"Journal of Electrocardiology"},{"key":"pcbi.1009862.ref015","unstructured":"Chaitanya K, Erdil E, Karani N, Konukoglu E. Contrastive Learning of Global and Local Features for Medical Image Segmentation with Limited Annotations. In: Larochelle H, Ranzato M, Hadsell R, Balcan MF, Lin H, editors. Advances in Neural Information Processing Systems. vol. 33. Curran Associates, Inc.; 2020. p. 12546\u201312558."},{"key":"pcbi.1009862.ref016","unstructured":"Zhang Y, Jiang H, Miura Y, Manning CD, Langlotz CP. Contrastive Learning of Medical Visual Representations from Paired Images and Text. arXiv:201000747 [cs]. 2020;."},{"key":"pcbi.1009862.ref017","doi-asserted-by":"crossref","unstructured":"Banville H, Albuquerque I, Hyvarinen A, Moffat G, Engemann DA, Gramfort A. Self-Supervised Representation Learning from Electroencephalography Signals. In: 2019 IEEE 29th International Workshop on Machine Learning for Signal Processing (MLSP). Pittsburgh, PA, USA: IEEE; 2019. p. 1\u20136.","DOI":"10.1109\/MLSP.2019.8918693"},{"key":"pcbi.1009862.ref018","unstructured":"Cheng JY, Goh H, Dogrusoz K, Tuzel O, Azemi E. Subject-Aware Contrastive Learning for Biosignals. arXiv:200704871 [cs, eess, stat]. 2020;."},{"key":"pcbi.1009862.ref019","doi-asserted-by":"crossref","first-page":"294","DOI":"10.1007\/978-3-319-67558-9_34","volume-title":"Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support","author":"A Jamaludin","year":"2017"},{"key":"pcbi.1009862.ref020","doi-asserted-by":"crossref","unstructured":"Chopra S, Hadsell R, LeCun Y. Learning a Similarity Metric Discriminatively, with Application to Face Verification. In: 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR\u201905). vol. 1; 2005. p. 539\u2013546 vol. 1.","DOI":"10.1109\/CVPR.2005.202"},{"key":"pcbi.1009862.ref021","unstructured":"Lin M, Chen Q, Yan S. Network in Network. In: Bengio Y, LeCun Y, editors. 2nd International Conference on Learning Representations, ICLR 2014, Banff, AB, Canada, April 14-16, 2014, Conference Track Proceedings; 2014."},{"key":"pcbi.1009862.ref022","unstructured":"Kingma DP, Ba J. Adam: A Method for Stochastic Optimization. In: Bengio Y, LeCun Y, editors. 3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Conference Track Proceedings; 2015."},{"key":"pcbi.1009862.ref023","unstructured":"Loshchilov I, Hutter F. SGDR: Stochastic Gradient Descent with Warm Restarts. In: 5th International Conference on Learning Representations, ICLR 2017, Toulon, France, April 24-26, 2017, Conference Track Proceedings. OpenReview.net; 2017."},{"key":"pcbi.1009862.ref024","doi-asserted-by":"crossref","unstructured":"Kolesnikov A, Zhai X, Beyer L. Revisiting Self-Supervised Visual Representation Learning. In: 2019 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Long Beach, CA, USA: IEEE; 2019. p. 1920\u20131929.","DOI":"10.1109\/CVPR.2019.00202"},{"key":"pcbi.1009862.ref025","unstructured":"Data Sciences Platform at Broad Institute of MIT and Harvard. ML4H; 2021. Available from: https:\/\/github.com\/broadinstitute\/ml4h."},{"issue":"8","key":"pcbi.1009862.ref026","doi-asserted-by":"crossref","first-page":"983","DOI":"10.1161\/01.STR.22.8.983","article-title":"Atrial Fibrillation as an Independent Risk Factor for Stroke: The Framingham Study","volume":"22","author":"PA Wolf","year":"1991","journal-title":"Stroke"},{"issue":"2","key":"pcbi.1009862.ref027","doi-asserted-by":"crossref","first-page":"417","DOI":"10.1016\/0735-1097(94)00371-V","article-title":"Electrocardiographic Identification of Increased Left Ventricular Mass by Simple Voltage-Duration Products","volume":"25","author":"PM Okin","year":"1995","journal-title":"Journal of the American College of Cardiology"},{"key":"pcbi.1009862.ref028","volume-title":"StatPearls","author":"AB Bornstein","year":"2021"},{"issue":"1","key":"pcbi.1009862.ref029","doi-asserted-by":"crossref","first-page":"18","DOI":"10.1049\/htl.2019.0096","article-title":"Detection and Classification of ECG Noises Using Decomposition on Mixed Codebook for Quality Analysis","volume":"7","author":"P Kumar","year":"2020","journal-title":"Healthcare Technology Letters"},{"key":"pcbi.1009862.ref030","doi-asserted-by":"crossref","first-page":"198","DOI":"10.1016\/j.cogsys.2018.07.004","article-title":"An Efficient Compression of ECG Signals Using Deep Convolutional Autoencoders","volume":"52","author":"O Yildirim","year":"2018","journal-title":"Cognitive Systems Research"},{"key":"pcbi.1009862.ref031","unstructured":"Ochiai K, Takahashi S. Arrhythmia Detection from 2-Lead ECG Using Convolutional Denoising Autoencoders. In: KDD\u201918 Deep Learning Day, London, UK; 2018."},{"key":"pcbi.1009862.ref032","doi-asserted-by":"crossref","unstructured":"Chen T, Guestrin C. XGBoost: A Scalable Tree Boosting System. In: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. KDD \u201916. New York, NY, USA: Association for Computing Machinery; 2016. p. 785\u2013794.","DOI":"10.1145\/2939672.2939785"},{"key":"pcbi.1009862.ref033","unstructured":"Tian Y, Sun C, Poole B, Krishnan D, Schmid C, Isola P. What Makes for Good Views for Contrastive Learning? In: Larochelle H, Ranzato M, Hadsell R, Balcan MF, Lin H, editors. Advances in Neural Information Processing Systems. vol. 33. Curran Associates, Inc.; 2020. p. 6827\u20136839."},{"issue":"23","key":"pcbi.1009862.ref034","doi-asserted-by":"crossref","first-page":"2851","DOI":"10.1001\/jama.291.23.2851","article-title":"Parental Atrial Fibrillation as a Risk Factor for Atrial Fibrillation in Offspring","volume":"291","author":"CS Fox","year":"2004","journal-title":"JAMA"}],"updated-by":[{"DOI":"10.1371\/journal.pcbi.1009862","type":"new_version","label":"New version","source":"publisher","updated":{"date-parts":[[2022,2,25]],"date-time":"2022-02-25T00:00:00Z","timestamp":1645747200000}}],"container-title":["PLOS Computational Biology"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pcbi.1009862","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,9,18]],"date-time":"2024-09-18T14:44:20Z","timestamp":1726670660000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pcbi.1009862"}},"subtitle":[],"editor":[{"given":"Roger Dimitri","family":"Kouyos","sequence":"first","affiliation":[]}],"short-title":[],"issued":{"date-parts":[[2022,2,14]]},"references-count":34,"journal-issue":{"issue":"2","published-online":{"date-parts":[[2022,2,14]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pcbi.1009862","relation":{"new_version":[{"id-type":"doi","id":"10.1371\/journal.pcbi.1009862","asserted-by":"object"}]},"ISSN":["1553-7358"],"issn-type":[{"value":"1553-7358","type":"electronic"}],"subject":[],"published":{"date-parts":[[2022,2,14]]}}}