{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,11,14]],"date-time":"2026-11-14T08:13:32Z","timestamp":1794644012844,"version":"4.9.1"},"reference-count":32,"publisher":"Oxford University Press (OUP)","issue":"3","license":[{"start":{"date-parts":[[2026,3,10]],"date-time":"2026-03-10T00:00:00Z","timestamp":1773100800000},"content-version":"vor","delay-in-days":6,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100003246","name":"Dutch Research Council","doi-asserted-by":"publisher","award":["628.011.213"],"award-info":[{"award-number":["628.011.213"]}],"id":[{"id":"10.13039\/501100003246","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001826","name":"Netherlands Organisation for Health Research and Development","doi-asserted-by":"publisher","award":["104021004"],"award-info":[{"award-number":["104021004"]}],"id":[{"id":"10.13039\/501100001826","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100002996","name":"Dutch Heart Foundation","doi-asserted-by":"crossref","award":["2019B011"],"award-info":[{"award-number":["2019B011"]}],"id":[{"id":"10.13039\/501100002996","id-type":"DOI","asserted-by":"crossref"}]},{"DOI":"10.13039\/501100002996","name":"Dutch Heart Foundation","doi-asserted-by":"crossref","award":["2022TKI2322"],"award-info":[{"award-number":["2022TKI2322"]}],"id":[{"id":"10.13039\/501100002996","id-type":"DOI","asserted-by":"crossref"}]},{"DOI":"10.13039\/100016036","name":"Health\u223cHolland","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100016036","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100016036","name":"Top Sector Life Sciences & Health","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100016036","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,3,4]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Aims<\/jats:title>\n                    <jats:p>Attribution-based explainability methods are widely used in electrocardiogram (ECG) analysis to interpret predictions from \u2018black-box\u2019 deep neural networks (DNNs). To be useful in clinical applications, attribution methods must produce explanations that are both clear and reflective of the model\u2019s inner workings. This study evaluates 12 attribution methods in DNN-based ECG classification.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods and results<\/jats:title>\n                    <jats:p>We analysed 12 attribution methods using a dataset of 873 710 median beat ECGs spanning nine diagnostic classes. Methods were applied to convolutional neural network-based models trained for ECG classification. Performance was evaluated across four experiments: inter-method similarity, self-consistency, dependence on model weights, and ability to identify features important for model inference. All task models achieved an area under the receiver operating curve above 0.95. Attribution methods demonstrated low correlation and high variability across inter-method comparisons. Self-consistency across random model initializations was moderate for most methods (mean correlation 0.41\u20130.65). Randomizing model weights led to rapid loss of correlation, although some methods did not converge to zero. Perturbation of input data revealed differences in how well attribution methods identified features relevant to model performance.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>Attribution methods demonstrated limited reliability, instability across model variants and incomplete dependence on learned parameters, constraining their utility in high-stakes settings such as healthcare. These findings suggest that attribution techniques should be used cautiously and supported by task-specific sanity checks. Approaches grounded in rigorous validation, inherently interpretable modelling or counterfactual explanations may better support clinically meaningful insight.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1093\/ehjdh\/ztag038","type":"journal-article","created":{"date-parts":[[2026,3,7]],"date-time":"2026-03-07T12:47:35Z","timestamp":1772887655000},"source":"Crossref","is-referenced-by-count":9,"title":["Signal or noise? Evaluating commonly used attribution methods for explaining deep neural networks in electrocardiogram classification"],"prefix":"10.1093","volume":"7","author":[{"ORCID":"https:\/\/orcid.org\/0009-0009-7494-4542","authenticated-orcid":false,"given":"Bauke K O","family":"Arends","sequence":"first","affiliation":[{"name":"Department of Cardiology, University Medical Center Utrecht , Internal ref E03.511, Heidelberglaan 100, 3584 CX Utrecht ,","place":["The Netherlands"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3181-0810","authenticated-orcid":false,"given":"Wouter A C","family":"van Amsterdam","sequence":"additional","affiliation":[{"name":"Department of Data Science and Biostatistics, Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht University , Utrecht ,","place":["The Netherlands"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2713-686X","authenticated-orcid":false,"given":"Pim","family":"van der Harst","sequence":"additional","affiliation":[{"name":"Department of Cardiology, University Medical Center Utrecht , Internal ref E03.511, Heidelberglaan 100, 3584 CX Utrecht ,","place":["The Netherlands"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5529-1541","authenticated-orcid":false,"given":"Maarten","family":"van Smeden","sequence":"additional","affiliation":[{"name":"Department of Data Science and Biostatistics, Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht University , Utrecht ,","place":["The Netherlands"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9950-4388","authenticated-orcid":false,"given":"Ren\u00e9","family":"van Es","sequence":"additional","affiliation":[{"name":"Department of Cardiology, University Medical Center Utrecht , Internal ref E03.511, Heidelberglaan 100, 3584 CX Utrecht ,","place":["The Netherlands"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4779-5870","authenticated-orcid":false,"given":"Rutger R","family":"van de Leur","sequence":"additional","affiliation":[{"name":"Department of Cardiology, University Medical Center Utrecht , Internal ref E03.511, Heidelberglaan 100, 3584 CX Utrecht ,","place":["The Netherlands"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"286","published-online":{"date-parts":[[2026,3,10]]},"reference":[{"key":"2026031208152096100_ztag038-B1","first-page":"1179","article-title":"Deep learning and the electrocardiogram: review of the current state-of-the-art","volume":"23","author":"Somani","year":"2021","journal-title":"EP Eur"},{"key":"2026031208152096100_ztag038-B2","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":"Hannun","year":"2019","journal-title":"Nat Med"},{"key":"2026031208152096100_ztag038-B3","doi-asserted-by":"crossref","first-page":"668","DOI":"10.1111\/jce.13889","article-title":"Prospective validation of a deep learning electrocardiogram algorithm for the detection of left ventricular systolic dysfunction","volume":"30","author":"Attia","year":"2019","journal-title":"J Cardiovasc Electrophysiol"},{"key":"2026031208152096100_ztag038-B4","doi-asserted-by":"crossref","first-page":"e745","DOI":"10.1016\/S2589-7500(21)00208-9","article-title":"The false hope of current approaches to explainable artificial intelligence in health care","volume":"3","author":"Ghassemi","year":"2021","journal-title":"Lancet Digit Health"},{"key":"2026031208152096100_ztag038-B5","doi-asserted-by":"crossref","first-page":"100686","DOI":"10.1016\/j.jacadv.2023.100686","article-title":"Clinical applications, methodology, and scientific reporting of electrocardiogram deep-learning models","volume":"2","author":"Avula","year":"2023","journal-title":"JACC Adv"},{"key":"2026031208152096100_ztag038-B6","doi-asserted-by":"crossref","first-page":"240","DOI":"10.1007\/s10462-024-10852-w","article-title":"A review of evaluation approaches for explainable AI with applications in cardiology","volume":"57","author":"Salih","year":"2024","journal-title":"Artif Intell Rev"},{"key":"2026031208152096100_ztag038-B7","doi-asserted-by":"crossref","first-page":"25","DOI":"10.1038\/s41746-024-01428-7","article-title":"Explainable AI associates ECG aging effects with increased cardiovascular risk in a longitudinal population study","volume":"8","author":"Hempel","year":"2025","journal-title":"Npj Digit Med"},{"key":"2026031208152096100_ztag038-B8","doi-asserted-by":"crossref","first-page":"e200267","DOI":"10.1148\/ryai.2021200267","article-title":"Assessing the trustworthiness of saliency maps for localizing abnormalities in medical imaging","volume":"3","author":"Arun","year":"2021","journal-title":"Radiol Artif Intell"},{"key":"2026031208152096100_ztag038-B9","volume-title":"Advances in Neural Information Processing Systems","author":"Adebayo","year":"2018"},{"key":"2026031208152096100_ztag038-B10","first-page":"1","volume-title":"2021 Digital Image Computing: Techniques and Applications (DICTA)","author":"Konate","year":"2021"},{"key":"2026031208152096100_ztag038-B11","doi-asserted-by":"crossref","first-page":"108525","DOI":"10.1016\/j.compbiomed.2024.108525","article-title":"Explaining deep learning for ECG analysis: building blocks for auditing and knowledge discovery","volume":"176","author":"Wagner","year":"2024","journal-title":"Comput Biol Med"},{"key":"2026031208152096100_ztag038-B12","doi-asserted-by":"crossref","first-page":"79","DOI":"10.1093\/jamia\/ocae280","article-title":"Evaluating gradient-based explanation methods for neural network ECG analysis using heatmaps","volume":"32","author":"Stor\u00e5s","year":"2024","journal-title":"J Am Med Inform Assoc"},{"key":"2026031208152096100_ztag038-B13","doi-asserted-by":"crossref","first-page":"109088","DOI":"10.1016\/j.compbiomed.2024.109088","article-title":"Visual interpretation of deep learning model in ECG classification: a comprehensive evaluation of feature attribution methods","volume":"182","author":"Suh","year":"2024","journal-title":"Comput Biol Med"},{"key":"2026031208152096100_ztag038-B14","doi-asserted-by":"crossref","first-page":"103655","DOI":"10.1016\/j.jbi.2020.103655","article-title":"The role of explainability in creating trustworthy artificial intelligence for health care: a comprehensive survey of the terminology, design choices, and evaluation strategies","volume":"113","author":"Markus","year":"2021","journal-title":"J Biomed Inform"},{"key":"2026031208152096100_ztag038-B15","doi-asserted-by":"crossref","first-page":"e015138","DOI":"10.1161\/JAHA.119.015138","article-title":"Automatic triage of 12-lead ECGs using deep convolutional neural networks","volume":"9","author":"Van De Leur","year":"2020","journal-title":"J Am Heart Assoc"},{"key":"2026031208152096100_ztag038-B16","doi-asserted-by":"publisher","author":"Vessies","DOI":"10.48550\/arXiv.2111.07386"},{"key":"2026031208152096100_ztag038-B17","doi-asserted-by":"publisher","author":"Simonyan","DOI":"10.48550\/arXiv.1312.6034"},{"key":"2026031208152096100_ztag038-B18","doi-asserted-by":"publisher","author":"Smilkov","DOI":"10.48550\/arXiv.1706.03825"},{"key":"2026031208152096100_ztag038-B19","doi-asserted-by":"publisher","author":"Shrikumar","DOI":"10.48550\/arXiv.1704.02685"},{"key":"2026031208152096100_ztag038-B20","doi-asserted-by":"publisher","author":"Springenberg","DOI":"10.48550\/arXiv.1412.6806"},{"key":"2026031208152096100_ztag038-B21","doi-asserted-by":"crossref","first-page":"618","DOI":"10.1109\/ICCV.2017.74","volume-title":"2017 IEEE International Conference on Computer Vision (ICCV)","author":"Selvaraju","year":"2017"},{"key":"2026031208152096100_ztag038-B22","doi-asserted-by":"crossref","first-page":"2921","DOI":"10.1109\/CVPR.2016.319","volume-title":"2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)","author":"Zhou","year":"2016"},{"key":"2026031208152096100_ztag038-B23","first-page":"1","volume-title":"Advances in Neural Information Processing Systems","author":"Lundberg","year":"2017"},{"key":"2026031208152096100_ztag038-B24","doi-asserted-by":"publisher","author":"Sundararajan","DOI":"10.48550\/arXiv.1703.01365"},{"key":"2026031208152096100_ztag038-B25","first-page":"1","article-title":"A consistent and ef\ufb01cient evaluation strategy for attribution methods","volume":"162","author":"Rong","year":"2022","journal-title":"Proc Mach Learn Res"},{"key":"2026031208152096100_ztag038-B26","doi-asserted-by":"crossref","first-page":"153792","DOI":"10.1016\/j.jelectrocard.2024.153792","article-title":"Clinical usability of deep learning-based saliency maps for occlusion myocardial infarction identification from the prehospital 12-lead electrocardiogram","volume":"87","author":"Riek","year":"2024","journal-title":"J Electrocardiol"},{"key":"2026031208152096100_ztag038-B27","doi-asserted-by":"crossref","first-page":"206","DOI":"10.1038\/s42256-019-0048-x","article-title":"Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead","volume":"1","author":"Rudin","year":"2019","journal-title":"Nat Mach Intell"},{"key":"2026031208152096100_ztag038-B28","doi-asserted-by":"crossref","first-page":"13122","DOI":"10.1038\/s41598-024-63656-x","article-title":"xECGArch: a trustworthy deep learning architecture for interpretable ECG analysis considering short-term and long-term features","volume":"14","author":"Goettling","year":"2024","journal-title":"Sci Rep"},{"key":"2026031208152096100_ztag038-B29","doi-asserted-by":"crossref","first-page":"23608","DOI":"10.1038\/s41598-025-08080-5","article-title":"A novel XAI framework for explainable AI-ECG using generative counterfactual XAI (GCX)","volume":"15","author":"Jang","year":"2025","journal-title":"Sci Rep"},{"key":"2026031208152096100_ztag038-B30","doi-asserted-by":"crossref","first-page":"680","DOI":"10.1093\/eurheartj\/ehac617","article-title":"Electrocardiogram-based deep learning improves outcome prediction following cardiac resynchronization therapy","volume":"44","author":"Wouters","year":"2023","journal-title":"Eur Heart J"},{"key":"2026031208152096100_ztag038-B31","first-page":"1","article-title":"ProtoECGNet: case-based interpretable deep learning for multi-label ECG classification with contrastive learning","volume":"298","author":"Sethi","year":"2025","journal-title":"Proc Mach Learn Res"},{"key":"2026031208152096100_ztag038-B32","doi-asserted-by":"crossref","first-page":"102009","DOI":"10.1016\/j.mex.2023.102009","article-title":"Generating post-hoc explanation from deep neural networks for multi-modal medical image analysis tasks","volume":"10","author":"Jin","year":"2023","journal-title":"MethodsX"}],"container-title":["European Heart Journal - Digital Health"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/ehjdh\/advance-article-pdf\/doi\/10.1093\/ehjdh\/ztag038\/67287781\/ztag038.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/ehjdh\/article-pdf\/7\/3\/ztag038\/67287781\/ztag038.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/ehjdh\/article-pdf\/7\/3\/ztag038\/67287781\/ztag038.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,12]],"date-time":"2026-03-12T12:15:31Z","timestamp":1773317731000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/ehjdh\/article\/doi\/10.1093\/ehjdh\/ztag038\/8512917"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,3,4]]},"references-count":32,"journal-issue":{"issue":"3","published-print":{"date-parts":[[2026,3,4]]}},"URL":"https:\/\/doi.org\/10.1093\/ehjdh\/ztag038","relation":{},"ISSN":["2634-3916"],"issn-type":[{"value":"2634-3916","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2026,4]]},"published":{"date-parts":[[2026,3,4]]},"article-number":"ztag038"}}