{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,15]],"date-time":"2026-10-15T07:52:25Z","timestamp":1792050745030,"version":"4.4.0"},"reference-count":27,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2025,2,17]],"date-time":"2025-02-17T00:00:00Z","timestamp":1739750400000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2025,2,17]],"date-time":"2025-02-17T00:00:00Z","timestamp":1739750400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"funder":[{"name":"Israel Ministry of Science and Technology"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["npj Digit. Med."],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>Despite the high accuracy of AI-based automated analysis of 12-lead ECG images for classification of cardiac conditions, clinical integration of such tools is hindered by limited interpretability of model recommendations. We aim to demonstrate the feasibility of a generic, clinical resource interpretability tool for AI models analyzing digitized 12-lead ECG images. To this end, we utilized the sensitivity of the Jacobian matrix to compute the gradient of the classifier for each pixel and provide medical relevance interpretability. Our methodology was validated using a dataset consisting of 79,226 labeled scanned ECG images, 11,316 unlabeled and 1807 labeled images obtained via mobile camera in clinical settings. The tool provided interpretability for both morphological and arrhythmogenic conditions, highlighting features in terms understandable to physician. It also emphasized significant signal features indicating the absence of certain cardiac conditions. High correlation was achieved between our method of interpretability and gold standard interpretations of 3 electrophysiologists.<\/jats:p>","DOI":"10.1038\/s41746-025-01467-8","type":"journal-article","created":{"date-parts":[[2025,2,17]],"date-time":"2025-02-17T23:01:28Z","timestamp":1739833288000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":31,"title":["Clinically meaningful interpretability of an AI model for ECG classification"],"prefix":"10.1038","volume":"8","author":[{"given":"Vadim","family":"Gliner","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Idan","family":"Levy","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5121-6974","authenticated-orcid":false,"given":"Kenta","family":"Tsutsui","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Moshe Rav","family":"Acha","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jorge","family":"Schliamser","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Assaf","family":"Schuster","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5183-6284","authenticated-orcid":false,"given":"Yael","family":"Yaniv","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2025,2,17]]},"reference":[{"key":"1467_CR1","first-page":"1","volume":"131","author":"V Gliner","year":"2023","unstructured":"Gliner, V., Makarov, V., Avetisyan, A. I., Schuster, A. & Yaniv, Y. Using domain adaptation for classification of healthy and disease conditions from mobile-captured images of standard 12-lead electrocardiograms. Sci. Reports131, 1\u201314 (2023).","journal-title":"Sci. Reports"},{"key":"1467_CR2","doi-asserted-by":"publisher","first-page":"80","DOI":"10.1016\/j.cvdhj.2023.03.002","volume":"4","author":"JCL Himmelreich","year":"2023","unstructured":"Himmelreich, J. C. L. & Harskamp, R. E. Diagnostic accuracy of the PMcardio smartphone application for artificial intelligence\u2013based interpretation of electrocardiograms in primary care (AMSTELHEART-1). Cardiovasc. Digit. Heal. J. 4, 80\u201390 (2023).","journal-title":"Cardiovasc. Digit. Heal. J."},{"key":"1467_CR3","doi-asserted-by":"crossref","unstructured":"Gliner, V. & Yaniv, Y. Identification of features for machine learning analysis for automatic arrhythmogenic event classification. Comput. in Cardiol. 44, 1\u20134 (2017).","DOI":"10.22489\/CinC.2017.170-101"},{"key":"1467_CR4","doi-asserted-by":"publisher","first-page":"88","DOI":"10.1016\/j.jelectrocard.2018.11.013","volume":"52","author":"SW Smith","year":"2019","unstructured":"Smith, S. W. et al. A deep neural network learning algorithm outperforms a conventional algorithm for emergency department electrocardiogram interpretation. J. Electrocardiol. 52, 88\u201395 (2019).","journal-title":"J. Electrocardiol."},{"key":"1467_CR5","first-page":"1","volume":"111","author":"AH Ribeiro","year":"2020","unstructured":"Ribeiro, A. H. et al. Automatic diagnosis of the 12-lead ECG using a deep neural network. Nat. Commun. 2020 111, 1\u20139 (2020).","journal-title":"Nat. Commun. 2020"},{"key":"1467_CR6","doi-asserted-by":"publisher","first-page":"1347","DOI":"10.1056\/NEJMra1814259","volume":"380","author":"A Rajkomar","year":"2019","unstructured":"Rajkomar, A., Dean, J. & Kohane, I. Machine Learning in Medicine. N. Engl. J. Med. 380, 1347\u20131358 (2019).","journal-title":"N. Engl. J. Med."},{"key":"1467_CR7","doi-asserted-by":"publisher","first-page":"30","DOI":"10.1038\/s41591-018-0307-0","volume":"25","author":"J He","year":"2019","unstructured":"He, J. et al. The practical implementation of artificial intelligence technologies in medicine. Nat. Med. 25, 30\u201336 (2019).","journal-title":"Nat. Med."},{"key":"1467_CR8","doi-asserted-by":"publisher","first-page":"468","DOI":"10.1016\/j.ijcard.2014.04.241","volume":"174","author":"JE Madias","year":"2014","unstructured":"Madias, J. E. Why the current diagnostic criteria of Takotsubo syndrome are outmoded: A proposal for new criteria. Int. J. Cardiol. 174, 468\u2013470 (2014).","journal-title":"Int. J. Cardiol."},{"key":"1467_CR9","doi-asserted-by":"crossref","unstructured":"Ayano, Y. M., Schwenker, F., Dufera, B. D. & Debelee, T. G. Interpretable Machine Learning Techniques in ECG-Based Heart Disease Classification: A Systematic Review. Diagnostics 13, 111 (2023).","DOI":"10.3390\/diagnostics13010111"},{"key":"1467_CR10","doi-asserted-by":"publisher","first-page":"104393","DOI":"10.1016\/j.compbiomed.2021.104393","volume":"133","author":"I Neves","year":"2021","unstructured":"Neves, I. et al. Interpretable heartbeat classification using local model-agnostic explanations on ECGs. Comput. Biol. Med. 133, 104393 (2021).","journal-title":"Comput. Biol. Med."},{"key":"1467_CR11","doi-asserted-by":"publisher","first-page":"1519","DOI":"10.1109\/JBHI.2020.3022989","volume":"25","author":"N Strodthoff","year":"2021","unstructured":"Strodthoff, N., Wagner, P., Schaeffter, T. & Samek, W. Deep Learning for ECG Analysis: Benchmarks and Insights from PTB-XL. IEEE J. Biomed. Heal. Informatics 25, 1519\u20131528 (2021).","journal-title":"IEEE J. Biomed. Heal. Informatics"},{"key":"1467_CR12","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-022-29153-3","volume":"13","author":"V Sangha","year":"2022","unstructured":"Sangha, V. et al. Automated multilabel diagnosis on electrocardiographic images and signals. Nat. Commun. 13, 1583 (2022).","journal-title":"Nat. Commun."},{"key":"1467_CR13","doi-asserted-by":"publisher","unstructured":"Mitra, R. N., Pramanik, S., Mitra, S. & Chaudhuri, B. B. A robust technique for delineation and features extraction of ECG signal from mobile-phone photography. in 2012 International Conference on Communications, Devices and Intelligent Systems (CODIS) 121\u2013124 https:\/\/doi.org\/10.1109\/CODIS.2012.6422151 (2012).","DOI":"10.1109\/CODIS.2012.6422151"},{"key":"1467_CR14","first-page":"2096","volume":"17","author":"Y Ganin","year":"2016","unstructured":"Ganin, Y. et al. Domain-adversarial training of neural networks. J. Mach. Learn. Res. Jmlr. 17, 2096 (2016).","journal-title":"J. Mach. Learn. Res. Jmlr."},{"key":"1467_CR15","doi-asserted-by":"publisher","unstructured":"Somwanshi, D., Tiwari, R., Saini, H. & Gupta, S. ECG Feature Extraction and Detection of First Degree Atrioventricular Block. in 2018 3rd International Conference and Workshops on Recent Advances and Innovations in Engineering (ICRAIE) 1\u20135 https:\/\/doi.org\/10.1109\/ICRAIE.2018.8710343 (2018).","DOI":"10.1109\/ICRAIE.2018.8710343"},{"key":"1467_CR16","doi-asserted-by":"publisher","first-page":"3348","DOI":"10.1109\/TBME.2012.2213597","volume":"59","author":"L Sun","year":"2012","unstructured":"Sun, L., Lu, Y., Yang, K. & Li, S. ECG Analysis Using Multiple Instance Learning for Myocardial Infarction Detection. IEEE Trans. Biomed. Eng. 59, 3348\u20133356 (2012).","journal-title":"IEEE Trans. Biomed. Eng."},{"key":"1467_CR17","doi-asserted-by":"publisher","first-page":"172","DOI":"10.15420\/aer.2021.26","volume":"10","author":"M Jastrz\u0119bski","year":"2021","unstructured":"Jastrz\u0119bski, M. ECG and Pacing Criteria for Differentiating Conduction System Pacing from Myocardial Pacing. Arrhythmia Electrophysiol. Rev. 10, 172\u2013180 (2021).","journal-title":"Arrhythmia Electrophysiol. Rev."},{"key":"1467_CR18","doi-asserted-by":"publisher","first-page":"116","DOI":"10.1016\/0002-8703(65)90224-3","volume":"69","author":"K-Y Chung","year":"1965","unstructured":"Chung, K.-Y., Walsh, T. J. & Massie, E. Wolff-Parkinson-White syndrome. Am. Heart J. 69, 116\u2013133 (1965).","journal-title":"Am. Heart J."},{"key":"1467_CR19","doi-asserted-by":"publisher","first-page":"E215","DOI":"10.1161\/01.CIR.101.23.e215","volume":"101","author":"AL Goldberger","year":"2000","unstructured":"Goldberger, A. L. et al. PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals. Circulation 101, E215\u2013E220 (2000).","journal-title":"Circulation"},{"key":"1467_CR20","first-page":"417","volume":"1323","author":"C Molnar","year":"2020","unstructured":"Molnar, C., Casalicchio, G. & Bischl, B. Interpretable Machine Learning \u2013 A Brief History, State-of-the-Art and Challenges. Commun. Comput. Inf. Sci. 1323, 417\u2013431 (2020).","journal-title":"Commun. Comput. Inf. Sci."},{"key":"1467_CR21","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.2020620118","volume":"118","author":"Y Elul","year":"2021","unstructured":"Elul, Y., Rosenberg, A. A., Schuster, A., Bronstein, A. M. & Yaniv, Y. Meeting the unmet needs of clinicians from AI systems showcased for cardiology with deep-learning\u2013based ECG analysis. Proc. Natl. Acad. Sci. USA 118, e2020620118 (2021).","journal-title":"Proc. Natl. Acad. Sci. USA"},{"key":"1467_CR22","doi-asserted-by":"publisher","first-page":"10","DOI":"10.1016\/j.tcm.2017.07.004","volume":"28","author":"OK Siddiqi","year":"2018","unstructured":"Siddiqi, O. K. & Ruberg, F. L. Cardiac amyloidosis: An update on pathophysiology, diagnosis, and treatment. Trends Cardiovasc. Med. 28, 10\u201321 (2018).","journal-title":"Trends Cardiovasc. Med."},{"key":"1467_CR23","doi-asserted-by":"publisher","first-page":"157","DOI":"10.1111\/anec.12018","volume":"18","author":"Z Cheng","year":"2013","unstructured":"Cheng, Z. et al. The findings of electrocardiography in patients with cardiac amyloidosis. Ann. Noninvasive ElectrocardiolJ.Int. Soc. Holter Noninvasive Electrocardiology, Inc. 18, 157\u2013162 (2013).","journal-title":"Ann. Noninvasive ElectrocardiolJ.Int. Soc. Holter Noninvasive Electrocardiology, Inc."},{"key":"1467_CR24","doi-asserted-by":"publisher","first-page":"286","DOI":"10.1016\/j.jelectrocard.2023.07.002","volume":"81","author":"KC Siontis","year":"2023","unstructured":"Siontis, K. C. et al. Saliency maps provide insights into artificial intelligence-based electrocardiography models for detecting hypertrophic cardiomyopathy. J. Electrocardiol. 81, 286\u2013291 (2023).","journal-title":"J. Electrocardiol."},{"key":"1467_CR25","doi-asserted-by":"publisher","unstructured":"Jin, L., Jin, M., Li, X. & Xu, X. A New Method Mobile Phone Image Denoising https:\/\/doi.org\/10.1117\/12.2204719 (2015).","DOI":"10.1117\/12.2204719"},{"key":"1467_CR26","doi-asserted-by":"crossref","unstructured":"Pech-Pacheco, J. L., Cristobal, G., Chamorro-Martinez, J. & Fernandez-Valdivia, J. Diatom autofocusing in brightfield microscopy: a comparative study. In Proc. 15th International Conference on Pattern Recognition. ICPR-2000 3, 314\u2013317 vol. 3 (2000).","DOI":"10.1109\/ICPR.2000.903548"},{"key":"1467_CR27","doi-asserted-by":"crossref","first-page":"58","DOI":"10.1063\/1.4902458","volume":"1631","author":"K Diederik","year":"2014","unstructured":"Diederik, K. & Ba, J. L. ADAM: A Method for Stochastic Optimization. AIP Conf. Proc. 1631, 58\u201362 (2014).","journal-title":"AIP Conf. Proc."}],"container-title":["npj Digital Medicine"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.nature.com\/articles\/s41746-025-01467-8.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41746-025-01467-8","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41746-025-01467-8.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,2,18]],"date-time":"2025-02-18T07:25:45Z","timestamp":1739863545000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.nature.com\/articles\/s41746-025-01467-8"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,2,17]]},"references-count":27,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2025,12]]}},"alternative-id":["1467"],"URL":"https:\/\/doi.org\/10.1038\/s41746-025-01467-8","relation":{},"ISSN":["2398-6352"],"issn-type":[{"value":"2398-6352","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,2,17]]},"assertion":[{"value":"7 July 2024","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"16 January 2025","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"17 February 2025","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"Vadim Gliner is an employee of Biosense Webster. Biosense Webster had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. All the other authors declare no competing interests.","order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Competing interests"}}],"article-number":"109"}}