{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,16]],"date-time":"2026-09-16T11:06:59Z","timestamp":1789556819556,"version":"build-2803163510"},"reference-count":30,"publisher":"BMJ","issue":"2","license":[{"start":{"date-parts":[[2026,7,16]],"date-time":"2026-07-16T00:00:00Z","timestamp":1784160000000},"content-version":"unspecified","delay-in-days":15,"URL":"https:\/\/creativecommons.org\/licenses\/by-nc\/4.0\/"}],"content-domain":{"domain":["bmj.com"],"crossmark-restriction":true},"short-container-title":["Open Heart"],"accepted":{"date-parts":[[2026,6,13]]},"published-print":{"date-parts":[[2026,7]]},"abstract":"<jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Differentiating heart failure (HF) with mildly reduced\/reduced ejection fraction (HFmr\/rEF) from HF with preserved ejection fraction (HFpEF) guides therapy but echocardiography may be delayed or unavailable. We developed and validated machine learning models using routine 12-lead ECG data to classify HF phenotypes.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>In this retrospective cohort of hospitalised patients with HF, predictors available at or before the index ECG were used. HFmrEF was pooled with HFrEF (left ventricular ejection fraction &lt;50%) for model development. Data were split 70\/30 into training and held-out test sets. Random forest (RF), Extreme Gradient Boosting and support vector machine models were trained and tuned using fivefold cross-validation in the training set. Boruta was used to select key ECG features. Test-set performance was evaluated by area under the curve (AUC) and accuracy; AUCs were compared using DeLong\u2019s test.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>Overall, 495 patients were included (254 HFmr\/rEF; 241 HFpEF). RF consistently performed best. Using ECG features alone, RF achieved an AUC of 0.821 (95% CI 0.752 to 0.890)\u2009and accuracy of 75.7%. A parsimonious RF model using 12 Boruta-selected ECG variables achieved an AUC of 0.832 (95% CI 0.766 to 0.899)\u2009and accuracy of 76.4%, with no significant AUC difference versus a comprehensive RF model using clinical\/laboratory\/ECG predictors (AUC 0.804, 95%\u2009CI 0.734 to 0.875; p=0.288) or the model using all ECG features (p=0.092). Adding X-ray cardiomegaly did not improve performance.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>A parsimonious RF model based on a small set of standard ECG measurements differentiates HFmr\/rEF from HFpEF with good discrimination, supporting ECG as an adjunct for phenotyping when echocardiography is not immediately available.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1136\/openhrt-2026-004127","type":"journal-article","created":{"date-parts":[[2026,7,16]],"date-time":"2026-07-16T15:25:50Z","timestamp":1784215550000},"page":"e004127","update-policy":"https:\/\/doi.org\/10.1136\/crossmarkpolicy","source":"Crossref","is-referenced-by-count":0,"title":["Differentiating HFmr\/rEF from HFpEF using standard 12-lead ECG measurements: an interpretable machine learning study"],"prefix":"10.1136","volume":"13","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-8201-9431","authenticated-orcid":false,"given":"Chenxu","family":"Zhao","sequence":"first","affiliation":[{"name":"Division of Cardiology, Department of Medicine and Therapeutics, Prince of Wales Hospital, Hong Kong, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0000-4809-9458","authenticated-orcid":false,"given":"Po Jie","family":"Chan","sequence":"additional","affiliation":[{"name":"The Chinese University of Hong Kong Faculty of Medicine, Hong Kong, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Scott","family":"Dougherty","sequence":"additional","affiliation":[{"name":"Division of Cardiology, Department of Medicine, Tseung Kwan O Hospital, Hong Kong, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4846-9823","authenticated-orcid":false,"given":"Yiting","family":"Fan","sequence":"additional","affiliation":[{"name":"Division of Electrocardiography and Echocardiography, Department of Cardiology, Shanghai General Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4120-155X","authenticated-orcid":false,"given":"Alex Pui Wai","family":"Lee","sequence":"additional","affiliation":[{"name":"Division of Cardiology, Department of Medicine and Therapeutics, Prince of Wales Hospital, Hong Kong, China"},{"name":"Li Ka Shing Institute of Health Sciences, The Chinese University of Hong Kong, Hong Kong, Hong Kong"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"239","published-online":{"date-parts":[[2026,7,16]]},"reference":[{"key":"2026091603501770000_13.2.e004127.1","doi-asserted-by":"publisher","DOI":"10.1016\/S0140-6736(18)32279-7"},{"key":"2026091603501770000_13.2.e004127.2","doi-asserted-by":"publisher","DOI":"10.1093\/cvr\/cvac013"},{"key":"2026091603501770000_13.2.e004127.3","doi-asserted-by":"publisher","DOI":"10.1093\/eurheartj\/ehab368"},{"key":"2026091603501770000_13.2.e004127.4","doi-asserted-by":"crossref","first-page":"30","DOI":"10.1016\/j.jelectrocard.2024.01.006","article-title":"Artificial intelligence-enhanced electrocardiography for accurate diagnosis and management of cardiovascular diseases","volume":"83","author":"Muzammil","year":"2024","journal-title":"J Electrocardiol"},{"key":"2026091603501770000_13.2.e004127.5","doi-asserted-by":"publisher","DOI":"10.1038\/s41591-022-02053-1"},{"key":"2026091603501770000_13.2.e004127.6","doi-asserted-by":"crossref","DOI":"10.1038\/s41746-023-00993-7","article-title":"Artificial intelligence-enabled ECG for left ventricular diastolic function and filling pressure","volume":"7","author":"Lee","year":"2024","journal-title":"NPJ Digit Med"},{"key":"2026091603501770000_13.2.e004127.7","doi-asserted-by":"crossref","first-page":"1044","DOI":"10.1093\/eurheartj\/ehae914","article-title":"Heart failure risk stratification using artificial intelligence applied to electrocardiogram images: a multinational study","volume":"46","author":"Dhingra","year":"2025","journal-title":"Eur Heart J"},{"key":"2026091603501770000_13.2.e004127.8","doi-asserted-by":"crossref","DOI":"10.1038\/s41598-023-33215-x","article-title":"Differences in learning characteristics between support vector machine and random forest models for compound classification revealed by Shapley value analysis","volume":"13","author":"Siemers","year":"2023","journal-title":"Sci Rep"},{"key":"2026091603501770000_13.2.e004127.9","doi-asserted-by":"crossref","DOI":"10.1038\/s41598-023-41093-6","article-title":"An efficient churn prediction model using gradient boosting machine and metaheuristic optimization","volume":"13","author":"AlShourbaji","year":"2023","journal-title":"Sci Rep"},{"key":"2026091603501770000_13.2.e004127.10","doi-asserted-by":"publisher","DOI":"10.1161\/cir.0000000000001063"},{"key":"2026091603501770000_13.2.e004127.11","doi-asserted-by":"publisher","DOI":"10.1002\/ejhf.813"},{"key":"2026091603501770000_13.2.e004127.12","doi-asserted-by":"publisher","DOI":"10.1038\/s41569-021-00605-5"},{"key":"2026091603501770000_13.2.e004127.13","doi-asserted-by":"crossref","first-page":"282","DOI":"10.1111\/cns.14002","article-title":"Automated machine learning\u2010based model for the prediction of delirium in patients after surgery for degenerative spinal disease","volume":"29","author":"Zhang","year":"2023","journal-title":"CNS Neurosci Ther"},{"key":"2026091603501770000_13.2.e004127.14","doi-asserted-by":"publisher","DOI":"10.3233\/FI-2010-288"},{"key":"2026091603501770000_13.2.e004127.15","doi-asserted-by":"publisher","DOI":"10.2307\/2531595"},{"key":"2026091603501770000_13.2.e004127.16","doi-asserted-by":"publisher","DOI":"10.1038\/s41591-018-0240-2"},{"key":"2026091603501770000_13.2.e004127.17","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-022-18640-8"},{"key":"2026091603501770000_13.2.e004127.18","doi-asserted-by":"crossref","first-page":"107283","DOI":"10.1109\/ACCESS.2022.3213065","article-title":"A New Medical Decision Support System for Diagnosing HFrEF and HFpEF Using ECG and Machine Learning Techniques","volume":"10","author":"Kavas","year":"2022","journal-title":"IEEE Access"},{"key":"2026091603501770000_13.2.e004127.19","doi-asserted-by":"crossref","first-page":"108357","DOI":"10.1016\/j.cmpb.2024.108357","article-title":"Multimodal deep learning models utilizing chest X-ray and electronic health record data for predictive screening of acute heart failure in emergency department","volume":"255","author":"Lee","year":"2024","journal-title":"Comput Methods Programs Biomed"},{"key":"2026091603501770000_13.2.e004127.20","doi-asserted-by":"crossref","first-page":"747","DOI":"10.1007\/s10278-024-01225-4","article-title":"A Deep-Learning-Enabled Electrocardiogram and Chest X-Ray for Detecting Pulmonary Arterial Hypertension","volume":"38","author":"Liu","year":"2025","journal-title":"J Imaging Inform Med"},{"key":"2026091603501770000_13.2.e004127.21","doi-asserted-by":"crossref","first-page":"108","DOI":"10.3390\/jcdd10030108","article-title":"Comparison of Depolarization and Repolarization Parameters in Left vs. Right Ventricular Septal Pacing\u2014An Intraprocedural Electrocardiographic Study","volume":"10","author":"Pestrea","year":"2023","journal-title":"JCDD"},{"key":"2026091603501770000_13.2.e004127.22","doi-asserted-by":"crossref","first-page":"71","DOI":"10.1016\/j.ijcard.2019.07.024","article-title":"Delayed prolongation of the QRS interval in patients with left ventricular dysfunction","volume":"296","author":"Rav-Acha","year":"2019","journal-title":"Int J Cardiol"},{"key":"2026091603501770000_13.2.e004127.23","doi-asserted-by":"crossref","first-page":"124","DOI":"10.1016\/j.jacep.2022.09.017","article-title":"Cardiac Repolarization in Health and Disease","volume":"9","author":"Krijger Ju\u00e1rez","year":"2023","journal-title":"JACC Clin Electrophysiol"},{"key":"2026091603501770000_13.2.e004127.24","doi-asserted-by":"crossref","DOI":"10.1038\/s41598-021-83137-9","article-title":"The Groningen electrocardiographic criteria for left ventricular hypertrophy: a sex-specific analysis","volume":"11","author":"van der Ende","year":"2021","journal-title":"Sci Rep"},{"key":"2026091603501770000_13.2.e004127.25","first-page":"100254","article-title":"XGBoost To Enhance Learner Performance Prediction","volume":"7","author":"Hakkal","year":"2024","journal-title":"Comput Educ: Artif Intell"},{"key":"2026091603501770000_13.2.e004127.26","doi-asserted-by":"crossref","DOI":"10.3390\/healthcare11121808","article-title":"Ensemble Learning for Disease Prediction: A Review","volume":"11","author":"Mahajan","year":"2023","journal-title":"Healthcare (Basel)"},{"key":"2026091603501770000_13.2.e004127.27","doi-asserted-by":"crossref","first-page":"1054","DOI":"10.1093\/eurheartj\/ehae906","article-title":"Using artificial intelligence to spot heart failure from ECGs: is it prime time?","volume":"46","author":"Antoniades","year":"2025","journal-title":"Eur Heart J"},{"key":"2026091603501770000_13.2.e004127.28","doi-asserted-by":"crossref","first-page":"1597","DOI":"10.1002\/ejhf.879","article-title":"Epidemiology, pathophysiology and clinical outcomes for heart failure patients with a mid-range ejection fraction","volume":"19","author":"Rastogi","year":"2017","journal-title":"Eur J Heart Fail"},{"key":"2026091603501770000_13.2.e004127.29","doi-asserted-by":"publisher","DOI":"10.1016\/j.jacc.2021.12.011"},{"key":"2026091603501770000_13.2.e004127.30","doi-asserted-by":"publisher","DOI":"10.1016\/S2589-7500(24)00172-9"}],"container-title":["Open Heart"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/syndication.highwire.org\/content\/doi\/10.1136\/openhrt-2026-004127","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,16]],"date-time":"2026-09-16T10:50:27Z","timestamp":1789555827000},"score":1,"resource":{"primary":{"URL":"https:\/\/openheart.bmj.com\/lookup\/doi\/10.1136\/openhrt-2026-004127"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7]]},"references-count":30,"journal-issue":{"issue":"2","published-online":{"date-parts":[[2026,7,16]]},"published-print":{"date-parts":[[2026,7]]}},"alternative-id":["10.1136\/openhrt-2026-004127"],"URL":"https:\/\/doi.org\/10.1136\/openhrt-2026-004127","relation":{},"ISSN":["2053-3624"],"issn-type":[{"value":"2053-3624","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7]]}}}