{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,11]],"date-time":"2026-07-11T03:25:24Z","timestamp":1783740324080,"version":"3.55.0"},"reference-count":53,"publisher":"JMIR Publications Inc.","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["J Med Internet Res"],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec sec-type=\"background\">\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Electrocardiogram (ECG) data constitutes one of the most widely available biosignal data in clinical and research settings, providing critical insights into cardiovascular diseases as well as broader health conditions. Advancements in deep learning demonstrate high performance in diverse ECG classification tasks, ranging from arrhythmia detection to risk prediction for various diseases. However, the widespread adoption of deep learning for ECG analysis faces significant barriers, including the heterogeneity of file formats, restricted access to pretrained model weights, and complex technical workflows for out-of-domain users.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec sec-type=\"objective\">\n                    <jats:title>Objective<\/jats:title>\n                    <jats:p>This study aims to address major bottlenecks in ECG-based deep learning by introducing ExChanGeAI, an open-source, web-based platform designed to offer an integrated, user-friendly platform for ECG data analysis. Our objective is to streamline the entire workflow\u2014from initial data ingestion (regardless of device or format) and intuitive visualization to privacy-preserving model training and task-specific fine-tuning\u2014making advanced ECG deep learning accessible for both clinical researchers and practitioners without machine learning (ML) expertise.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec sec-type=\"methods\">\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>ExChanGeAI incorporates robust preprocessing modules for various ECG file types, a set of interactive visualization tools for exploratory data analysis, and multiple state-of-the-art deep learning architectures for ECGs. Users can choose to train models from scratch or fine-tune pretrained models using their own datasets, while all computations are performed locally to ensure data privacy. The platform is adaptable for deployment on personal computers as well as scalable to high-performance computing infrastructures. We demonstrate the platform\u2019s performance on several clinically relevant classification tasks across 3 external and heterogeneous validation datasets, including a newly curated test set from routine care, evaluating both model generalizability and resource efficiency.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec sec-type=\"results\">\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>Our experiments show that de novo training with user-provided, task-specific data can outperform a leading foundation model, while requiring substantially fewer parameters and computational resources. The platform enables users to empirically determine the most suitable model for their specific tasks, based on systematic validations, while lowering technical barriers for out-of-domain experts and promoting open research.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec sec-type=\"conclusions\">\n                    <jats:title>Conclusions<\/jats:title>\n                    <jats:p>ExChanGeAI provides a comprehensive, privacy-aware platform that democratizes access to ECG analysis and model training. By simplifying complex workflows, ExChanGeAI empowers out-of-domain researchers to use state-of-the-art ML on diverse datasets, democratizing the access to ML in the field of ECG data. The platform is available as open-source code under the Massachusetts Institute of Technology (MIT) license.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.2196\/81116","type":"journal-article","created":{"date-parts":[[2026,1,30]],"date-time":"2026-01-30T20:55:05Z","timestamp":1769806505000},"page":"e81116-e81116","source":"Crossref","is-referenced-by-count":1,"title":["End-to-End Platform for Electrocardiogram Analysis and Model Fine-Tuning: Development and Validation Study"],"prefix":"10.2196","volume":"28","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-9043-7844","authenticated-orcid":false,"given":"Lucas","family":"Bickmann","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7626-8853","authenticated-orcid":false,"given":"Lucas","family":"Plagwitz","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1774-4252","authenticated-orcid":false,"given":"Antonius","family":"B\u00fcscher","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3735-8503","authenticated-orcid":false,"given":"Lars","family":"Eckardt","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7206-3719","authenticated-orcid":false,"given":"Julian","family":"Varghese","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1010","published-online":{"date-parts":[[2026,1,30]]},"reference":[{"issue":"5","key":"R1","doi-asserted-by":"publisher","first-page":"1461","DOI":"10.1038\/s41591-024-02961-4","article-title":"AI-enabled electrocardiography alert intervention and all-cause mortality: a pragmatic randomized clinical trial","volume":"30","author":"Lin","journal-title":"Nat Med"},{"issue":"10","key":"R2","doi-asserted-by":"publisher","first-page":"2897","DOI":"10.1038\/s41591-024-03243-9","article-title":"Artificial intelligence guided screening for cardiomyopathies in an obstetric population: a pragmatic randomized clinical trial","volume":"30","author":"Adedinsewo","journal-title":"Nat Med"},{"issue":"11","key":"R3","doi-asserted-by":"publisher","first-page":"e791","DOI":"10.1016\/S2589-7500(24)00172-9","article-title":"Artificial intelligence-enabled electrocardiogram for mortality and cardiovascular risk estimation: a model development and validation study","volume":"6","author":"Sau","journal-title":"Lancet Digit Health"},{"issue":"8040","key":"R4","doi-asserted-by":"publisher","first-page":"827","DOI":"10.1038\/s41586-024-08141-1","article-title":"Why \u201copen\u201d AI systems are actually closed, and why this matters","volume":"635","author":"Widder","journal-title":"Nature New Biol"},{"key":"R5","doi-asserted-by":"publisher","DOI":"10.1016\/j.cca.2023.117519","article-title":"Advancing AI in healthcare: a comprehensive review of best practices","volume":"548","author":"Polevikov","journal-title":"Clin Chim Acta"},{"key":"R6","doi-asserted-by":"publisher","DOI":"10.1177\/2632010X241226887","article-title":"Impact of artificial intelligence (AI) technology in the healthcare sector: a critical evaluation of both sides of the coin","volume":"17","author":"Rahman","journal-title":"Clin Med Insights Pathol"},{"issue":"5","key":"R7","doi-asserted-by":"publisher","first-page":"7374","DOI":"10.1109\/JIOT.2023.3329061","article-title":"Federated learning for medical applications: a taxonomy, current trends, challenges, and future research directions","volume":"11","author":"Rauniyar","journal-title":"IEEE Internet Things J"},{"issue":"1","key":"R8","doi-asserted-by":"publisher","first-page":"19","DOI":"10.4258\/hir.2021.27.1.19","article-title":"Effectiveness of transfer learning for deep learning-based electrocardiogram analysis","volume":"27","author":"Jang","journal-title":"Healthc Inform Res"},{"issue":"1","key":"R9","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-021-84374-8","article-title":"Transfer learning for ECG classification","volume":"11","author":"Weimann","journal-title":"Sci Rep"},{"issue":"12","key":"R10","doi-asserted-by":"publisher","DOI":"10.3390\/jpm13121703","article-title":"Survey of transfer learning approaches in the machine learning of digital health sensing data","volume":"13","author":"Chato","journal-title":"J Pers Med"},{"key":"R11","doi-asserted-by":"publisher","DOI":"10.1016\/j.bspc.2024.106160","article-title":"Deep neural networks generalization and fine-tuning for 12-lead ECG classification","volume":"93","author":"Avetisyan","journal-title":"Biomed Signal Process Control"},{"key":"R12","doi-asserted-by":"publisher","DOI":"10.1016\/j.bspc.2018.03.003","article-title":"A survey on ECG analysis","volume":"43","author":"Kaplan Berkaya","journal-title":"Biomed Signal Process Control"},{"key":"R13","doi-asserted-by":"publisher","DOI":"10.1016\/j.eswax.2020.100033","article-title":"A review on deep learning methods for ECG arrhythmia classification","volume":"7","author":"Ebrahimi","journal-title":"Expert Syst Appl: X"},{"key":"R14","doi-asserted-by":"publisher","DOI":"10.1016\/j.bspc.2023.105780","article-title":"A lightweight SelfONN model for general ECG classification with pretraining","volume":"89","author":"Qin","journal-title":"Biomed Signal Process Control"},{"issue":"1","key":"R15","doi-asserted-by":"publisher","first-page":"12823","DOI":"10.1038\/s41598-024-63378-0","article-title":"ECG autoencoder based on low-rank attention","volume":"14","author":"Zhang","journal-title":"Sci Rep"},{"key":"R16","doi-asserted-by":"publisher","DOI":"10.1016\/j.engappai.2023.106484","article-title":"ECG-NET: a deep LSTM autoencoder for detecting anomalous ECG","volume":"124","author":"Roy","journal-title":"Eng Appl Artif Intell"},{"issue":"1","key":"R17","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-021-01295-2","article-title":"DeepFake electrocardiograms using generative adversarial networks are the beginning of the end for privacy issues in medicine","volume":"11","author":"Thambawita","journal-title":"Sci Rep"},{"key":"R18","doi-asserted-by":"publisher","DOI":"10.1109\/TIM.2022.3197757","article-title":"Attention-based convolutional denoising autoencoder for two-lead ECG denoising and arrhythmia classification","volume":"71","author":"Singh","journal-title":"IEEE Trans Instrum Meas"},{"issue":"23","key":"R19","doi-asserted-by":"publisher","first-page":"e215","DOI":"10.1161\/01.CIR.101.23.e215","article-title":"PhysioBank, PhysioToolkit, and PhysioNet","volume":"101","author":"Goldberger","journal-title":"Circulation"},{"issue":"1","key":"R20","doi-asserted-by":"publisher","DOI":"10.1038\/s41597-020-0495-6","article-title":"PTB-XL, a large publicly available electrocardiography dataset","volume":"7","author":"Wagner","journal-title":"Sci Data"},{"key":"R21","doi-asserted-by":"publisher","DOI":"10.3233\/SHTI230099","article-title":"Post Hoc sample size estimation for deep learning architectures for ECG-classification","volume":"302","author":"Bickmann","journal-title":"Stud Health Technol Inform"},{"issue":"6","key":"R22","doi-asserted-by":"publisher","first-page":"1936","DOI":"10.1007\/s10618-020-00710-y","article-title":"InceptionTime: finding AlexNet for time series classification","volume":"34","author":"Ismail Fawaz","journal-title":"Data Min Knowl Disc"},{"key":"R23","doi-asserted-by":"crossref","unstructured":"Rahimian E Zabihi S Atashzar SF Asif A Mohammadi A . XceptionTime: independent time-window xceptiontime architecture for hand gesture classification. Presented at: ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP); May 4-8, 2020:1304-1308; Barcelona, Spain. [doi: 10.1109\/ICASSP40776.2020.9054586]","DOI":"10.1109\/ICASSP40776.2020.9054586"},{"key":"R24","unstructured":"Han Y Murino V Liu X Zhang X Ding C . A systematic review on foundation models for electrocardiogram analysis: initial strides and expansive horizons. arXiv. Preprint posted online on Oct 13, 2025. [doi: 10.48550\/arXiv.2410.19877]"},{"key":"R25","unstructured":"Li J Aguirre A Moura J et al. An electrocardiogram foundation model built on over 10 million recordings with external evaluation across multiple domains. arXiv. Preprint posted online on Aug 4, 2025. [doi: 10.48550\/arXiv.2410.04133]"},{"key":"R26","unstructured":"Wang Y Cao X Hu Y et al. AnyECG: foundational models for multitask cardiac analysis in real-world settings. arXiv. Preprint posted online on Mar 3, 2025. [doi: 10.48550\/arXiv.2411.17711]"},{"key":"R27","doi-asserted-by":"publisher","DOI":"10.1016\/j.inffus.2025.103363","article-title":"ECGFM: a foundation model for ECG analysis trained on a multi-center million-ECG dataset","volume":"124","author":"Zhang","journal-title":"Inf Fusion"},{"issue":"5","key":"R28","doi-asserted-by":"publisher","DOI":"10.1093\/jamiaopen\/ooaf122","article-title":"ECG-FM: an open electrocardiogram foundation model","volume":"8","author":"McKeen","journal-title":"JAMIA Open"},{"issue":"1","key":"R29","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1038\/s44325-024-00027-5","article-title":"Foundation models for cardiovascular disease detection via biosignals from digital stethoscopes","volume":"1","author":"Mathew","journal-title":"npj Cardiovasc Health"},{"issue":"1","key":"R30","doi-asserted-by":"publisher","first-page":"2349","DOI":"10.5555\/2567709.2567736","article-title":"Orange: data mining toolbox in python","volume":"14","author":"Dem\u0161ar","journal-title":"J Mach Learn Res"},{"issue":"33","key":"R31","doi-asserted-by":"publisher","first-page":"1132","DOI":"10.21105\/joss.01132","article-title":"GAMA: Genetic automated machine learning assistant","volume":"4","author":"Gijsbers","journal-title":"J Open Source Softw"},{"key":"R32","unstructured":"Bickmann L . ExChanGeAI. GitHub. URL: https:\/\/github.com\/VargheseLab\/exchangeai [Accessed 24-12-2025]"},{"issue":"4","key":"R33","doi-asserted-by":"publisher","first-page":"1689","DOI":"10.3758\/s13428-020-01516-y","article-title":"NeuroKit2: a Python toolbox for neurophysiological signal processing","volume":"53","author":"Makowski","journal-title":"Behav Res Methods"},{"key":"R34","unstructured":"Plagwitz L Bickmann L Fujarski M et al. The Rlign algorithm for enhanced electrocardiogram analysis through R-peak alignment for explainable classification and clustering. arXiv. Preprint posted online on Aug 9, 2024. [doi: 10.48550\/arXiv.2407.15555]"},{"key":"R35","unstructured":"Salimi A Kalmady SV Hindle A Zaiane O Kaul P . Exploring best practices for ECG signal processing in machine learning. arXiv. Preprint posted online on May 14, 2025. [doi: 10.48550\/arXiv.2311.04229]"},{"issue":"11","key":"R36","doi-asserted-by":"publisher","first-page":"5326","DOI":"10.1109\/JBHI.2023.3310989","article-title":"Towards quantitative precision for ECG analysis: leveraging state space models, self-supervision and patient metadata","volume":"27","author":"Mehari","journal-title":"IEEE J Biomed Health Inform"},{"issue":"5","key":"R37","doi-asserted-by":"publisher","DOI":"10.3390\/s22051776","article-title":"Compressed deep learning to classify arrhythmia in an embedded wearable device","volume":"22","author":"Lee","journal-title":"Sensors (Basel)"},{"key":"R38","doi-asserted-by":"publisher","DOI":"10.1093\/eurheartj\/ehaf254","article-title":"Deep learning electrocardiogram model for risk stratification of coronary revascularization need in the emergency department","author":"B\u00fcscher","journal-title":"Eur Heart J"},{"key":"R39","doi-asserted-by":"crossref","unstructured":"Reyna MA Sadr N Alday EAP et al. Will two do? varying dimensions in electrocardiography: the physionet\/computing in cardiology challenge 2021. Presented at: 2021 Computing in Cardiology (CinC); Sep 13-15, 2021:1-4; Brno, Czech Republic. [doi: 10.23919\/CinC53138.2021.9662687]","DOI":"10.23919\/CinC53138.2021.9662687"},{"key":"R40","doi-asserted-by":"crossref","unstructured":"Han H Park S Min S et al. Towards high generalization performance on electrocardiogram classification. Presented at: 2021 Computing in Cardiology (CinC); Sep 13-15, 2021:1-4; Brno, Czech Republic. [doi: 10.23919\/CinC53138.2021.9662737]","DOI":"10.23919\/CinC53138.2021.9662737"},{"key":"R41","doi-asserted-by":"crossref","unstructured":"Ansel J Yang E He H et al. PyTorch 2: faster machine learning through dynamic python bytecode transformation and graph compilation. Presented at: Proceedings of the 29th ACM International Conference on Architectural Support for Programming Languages and Operating Systems; Apr 27, 2024; La Jolla, CA. Apr 27, 2024.[doi: 10.1145\/3620665.3640366]","DOI":"10.1145\/3620665.3640366"},{"key":"R42","unstructured":"Loshchilov I Hutter F . Decoupled Weight Decay Regularization. Presented at: International Conference on Learning Representations; May 6-9, 2019; New Orleans, Louisiana, United States. 2019.URL: https:\/\/openreview.net\/forum?id=Bkg6RiCqY7 [Accessed 28-01-2026]"},{"issue":"9","key":"R43","doi-asserted-by":"publisher","first-page":"6486","DOI":"10.1109\/TPAMI.2024.3382294","article-title":"Towards understanding convergence and generalization of AdamW","volume":"46","author":"Zhou","journal-title":"IEEE Trans Pattern Anal Mach Intell"},{"key":"R44","unstructured":"Kingma DP Ba J . Adam: a method for stochastic optimization. arXiv. Preprint posted online on Jan 30, 2017. URL: https:\/\/arxiv.org\/abs\/1412.6980 [Accessed 24-12-2025] [doi: 10.48550\/arXiv.1412.6980]"},{"key":"R45","doi-asserted-by":"crossref","unstructured":"Li J Yang X . A cyclical learning rate method in deep learning training. Presented at: 2020 International Conference on Computer, Information and Telecommunication Systems (CITS); Oct 5-7, 2020:1-5; Hangzhou, China. [doi: 10.1109\/CITS49457.2020.9232482]","DOI":"10.1109\/CITS49457.2020.9232482"},{"issue":"5","key":"R46","doi-asserted-by":"publisher","first-page":"1519","DOI":"10.1109\/JBHI.2020.3022989","article-title":"Deep learning for ECG analysis: benchmarks and insights from PTB-XL","volume":"25","author":"Strodthoff","journal-title":"IEEE J Biomed Health Inform"},{"issue":"1","key":"R47","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-023-39472-8","article-title":"Practical intelligent diagnostic algorithm for wearable 12-lead ECG via self-supervised learning on large-scale dataset","volume":"14","author":"Lai","journal-title":"Nat Commun"},{"key":"R48","unstructured":"Yang W Feng Q . Offline test set of ECG multi-label classfication. Science Data Bank; 2023. [doi: 10.57760\/sciencedb.07677]"},{"key":"R49","unstructured":"Gow B Pollard T Nathanson LA et al. MIMIC-IV-ECG: diagnostic electrocardiogram matched subset. PhysioNet; [doi: 10.13026\/4nqg-sb35]"},{"issue":"5","key":"R50","doi-asserted-by":"publisher","first-page":"845","DOI":"10.1109\/TNNLS.2013.2292894","article-title":"Classification in the presence of label noise: a survey","volume":"25","author":"Fr\u00e9nay","journal-title":"IEEE Trans Neural Netw Learn Syst"},{"issue":"8","key":"R51","doi-asserted-by":"publisher","first-page":"084001","DOI":"10.1088\/1361-6579\/ac79fd","article-title":"Issues in the automated classification of multilead ecgs using heterogeneous labels and populations","volume":"43","author":"Reyna","journal-title":"Physiol Meas"},{"issue":"4","key":"R52","doi-asserted-by":"publisher","DOI":"10.3390\/jcdd10040175","article-title":"Current and future use of artificial intelligence in electrocardiography","volume":"10","author":"Mart\u00ednez-Sell\u00e9s","journal-title":"J Cardiovasc Dev Dis"},{"issue":"586","key":"R53","doi-asserted-by":"publisher","DOI":"10.1126\/scitranslmed.abb1655","article-title":"Reproducibility in machine learning for health research: still a ways to go","volume":"13","author":"McDermott","journal-title":"Sci Transl Med"}],"container-title":["Journal of Medical Internet Research"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.jmir.org\/2026\/1\/e81116","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,30]],"date-time":"2026-06-30T18:14:49Z","timestamp":1782843289000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.jmir.org\/2026\/1\/e81116"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,1,30]]},"references-count":53,"URL":"https:\/\/doi.org\/10.2196\/81116","relation":{},"ISSN":["1438-8871"],"issn-type":[{"value":"1438-8871","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,1,30]]},"article-number":"v28i5e81116"}}