{ "abstract": "Contrastive learning is a widely adopted self-supervised pretraining strategy, yet its dependence on cohort composition remains underexplored. We present Contrasting by Augmented Patient Electrocardiograms (CAPE) foundation model and pretrain on four cohorts (n = 5,203,269), from diverse populations across three continents (North America, South America, Asia). We systematically assess how cohort demographics, health status, and population diversity influence the downstream performance for prediction tasks also including two additional cohorts from another continent (Europe). We find that downstream performance depends on the distributional properties of the pretraining cohort, including demographics and health status. Moreover, while pretraining with a multi-centre, demographically diverse cohort improves in-distribution accuracy, it reduces out-of-distribution (OOD) generalisation of our contrastive approach by encoding cohort-specific artifacts. To address this, we propose the In-Distribution Batch (IDB) strategy, which preserves intra-cohort consistency during pretraining, discourages learning of spurious cohort-specific features, and instead promotes clinically meaningful variability within cohorts. This leads to improved out-of-distribution robustness, with gains of 9–40% in downstream label prediction performance. This work provides insights into pretraining strategies for more clinically deployable and generalisable foundation models. Author summary: Artificial intelligence (AI) can learn from large collections of electrocardiograms (ECGs) and support the development of new diagnostic tools without relying on extensive manual annotation. However, it remains unclear how the data used to pretrain these models influences their ability to generalise across different patient populations.", "abstract_shown": true, "affiliations": [ "National Heart and Lung Institute, Imperial College London, London, United Kingdom", "Department of Cardiology, Chelsea and Westminster Hospital NHS Foundation Trust, London, United Kingdom", "Department of Cardiology, Zhongshan Hospital of Fudan University, Shanghai Institute of Cardiovascular Diseases, National Clinical Research Centre for Interventional Medicine, Shanghai, China", "Department of Information Technology, Uppsala University, Uppsala, Sweden", "Vanderbilt Institute for Clinical and Translational Research, Vanderbilt University Medical Center, Nashville, Tennessee, United States of America", "Center for Digital Genomic Medicine, Department of Medicine, Vanderbilt University Medical Center, Nashville, Tennessee, United States of America", "Division of Cardiovascular Medicine, Vanderbilt University Medical Center, Nashville, Tennessee, United States of America", "Department of Internal Medicine, Faculdade de Medicina, and Telehealth Center and Cardiology Service, Hospital das Clínicas, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil", "Richard A. and Susan F. Smith Center for Outcomes Research in Cardiology, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts, United States of America", "Harvard-Thorndike Electrophysiology Institute, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts, United States of America", "Department of Cardiology, Imperial College Healthcare NHS Trust, London, United Kingdom" ], "article_type": "Research Article", "arxiv": null, "author_total": 20, "authors": [ { "affs": [ 0 ], "name": "Gul Rukh Khattak" }, { "affs": [ 1 ], "name": "Konstantinos Patlatzoglou" }, { "affs": [ 2 ], "name": "Joseph Barker" }, { "affs": [ 3 ], "name": "Libor Pastika" }, { "affs": [ 4 ], "name": "Boroumand Zeidaabadi" }, { "affs": [ 5 ], "name": "Aidan R. Birdi" }, { "affs": [ 6 ], "name": "Jiayu Huo" }, { "affs": [ 7 ], "name": "Ahmed El-Medany" }, { "affs": [ 8 ], "name": "Hesham Aggour" }, { "affs": [ 9 ], "name": "Yixiu Liang" }, { "affs": [ 10 ], "name": "Antonio H. Ribeiro" }, { "name": "Jeffrey Annis" }, { "name": "Antonio Luiz Pinho Ribeiro" }, { "name": "Junbo Ge" }, { "name": "Daniel B. Kramer" }, { "name": "Jonathan W. Waks" }, { "name": "Evan Brittain" }, { "name": "Nicholas Peters" }, { "name": "Fu Siong Ng" }, { "name": "Arunashis Sau" } ], "bibtex": "https://research.draconitic.com/paper/6n2dbbka9pefh.bib", "cited_by": 1, "doi": "10.1371/journal.pdig.0001623", "fulltext_url": "https://journals.plos.org/digitalhealth/article?id=10.1371/journal.pdig.0001623", "id": "6n2dbbka9pefh", "issns": [ "2767-3170" ], "landing_url": "https://journals.plos.org/digitalhealth/article?id=10.1371/journal.pdig.0001623", "language": null, "licence": "CC-BY-4.0", "licence_label": "CC BY 4.0", "licence_source": "journal", "licence_url": "https://creativecommons.org/licenses/by/4.0/", "metadata_licence": null, "native_subjects": [ [ "openalex", "2705|ECG Monitoring and Analysis" ], [ "plos", "/Biology and life sciences/Ecology/Ecological metrics/Species diversity" ], [ "plos", "/Biology and life sciences/Neuroscience/Cognitive science/Cognitive psychology/Learning" ], [ "plos", "/Biology and life sciences/Neuroscience/Learning and memory/Learning" ], [ "plos", "/Biology and life sciences/Population biology/Population metrics/Age distribution" ], [ "plos", "/Biology and life sciences/Psychology/Cognitive psychology/Learning" ], [ "plos", "/Computer and information sciences/Artificial intelligence/Machine learning" ], [ "plos", "/Computer and information sciences/Artificial intelligence/Machine learning/Machine learning algorithms/Perceptrons/Multilayer perceptrons" ], [ "plos", "/Computer and information sciences/Software engineering/Preprocessing" ], [ "plos", "/Ecology and environmental sciences/Ecology/Ecological metrics/Species diversity" ], [ "plos", "/Engineering and technology/Software engineering/Preprocessing" ], [ "plos", "/Physical sciences/Mathematics/Applied mathematics/Algorithms/Machine learning algorithms/Perceptrons/Multilayer perceptrons" ], [ "plos", "/Physical sciences/Mathematics/Probability theory/Statistical distributions" ], [ "plos", "/Research and analysis methods/Bioassays and physiological analysis/Electrophysiological techniques/Cardiac electrophysiology/Electrocardiography" ], [ "plos", "/Research and analysis methods/Simulation and modeling/Algorithms/Machine learning algorithms/Perceptrons/Multilayer perceptrons" ], [ "plos", "/Social sciences/Psychology/Cognitive psychology/Learning" ] ], "openalex": "W4414596658", "pdf_url": "https://journals.plos.org/digitalhealth/article/file?id=10.1371/journal.pdig.0001623&type=printable", "pmcid": null, "pmid": null, "published": "2026-09-02", "publisher": "Public Library of Science", "ris": "https://research.draconitic.com/paper/6n2dbbka9pefh.ris", "sources": [ { "first_seen": "2026-10-30T12:18:02Z", "native_id": "10.1371/journal.pdig.0001623", "source": "plos", "url": "https://journals.plos.org/digitalhealth/article?id=10.1371/journal.pdig.0001623" } ], "subjects": [ { "field": "biology", "origin": "plos:ecology", "subfield": "ecology" }, { "field": "biology", "origin": "plos:neuroscience", "subfield": "neuroscience" }, { "field": "mathematics", "origin": "plos:probability theory", "subfield": "probability" } ], "title": "Data distribution impacts the performance and generalisability of contrastive learning-based foundation models of electrocardiograms", "title_text": "Data distribution impacts the performance and generalisability of contrastive learning-based foundation models of electrocardiograms", "type": "peer-reviewed", "updated_at": "2026-10-30T12:18:02Z", "url": "https://research.draconitic.com/paper/6n2dbbka9pefh", "venue": "PLOS Digital Health", "versions": [] }