{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,14]],"date-time":"2026-09-14T06:04:07Z","timestamp":1789365847046,"version":"build-2803163510"},"reference-count":32,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,9,14]],"date-time":"2026-09-14T00:00:00Z","timestamp":1789344000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Artif. Intell."],"abstract":"<jats:p>Artificial Intelligence (AI) holds the promise of transforming patient care in healthcare, but the general adoption of AI in the healthcare sector is impeded by issues that stem from a lack of transparency and trust in AI models. We noted a lack of faith in artificial intelligence (AI) systems working together with cardiologists, which can be considered \u201cblack-boxes. This observation led to the idea of incorporating interpretability into the architecture itself, which is achieved by the key innovation of multi-scale feature abstraction and hierarchical attention mechanisms. A novel architecture, namely MSHAN (Multi-Scale Hybrid Attention Network), is proposed to address the critical challenge with an overall accuracy of 96.25% for ECG arrhythmia detection. The MIT-BIH Arrhythmia Database was used for experiments. The proposed model exhibited good discrimination and classification performance for the inter-patient DS2 test set with macro-averaged sensitivity more than 95.34%, macro-averaged precision more than 95.01%, and specificity more than 98.36% for all the classes of arrhythmia, and the discriminative capability was also observed by probability-based ROC analysis, with an average AUC of 0.980. The results of the comparative analysis showed the superiority of the MSHAN method in comparison with traditional methods, including the ResNet-1D (92.14%) and 1D CNN (88.73%), and the explanation of the predictions made by the MSHAN for the inter-patient evaluation mode. We performed an informal clinical observation that the visualizations of the model\u2019s decision process, based on SHAP, helped the cardiologists in better understanding and checking the correctness of the model\u2019s predictions; formal clinical check and validation are topics that will be explored in future. A wide range of experiments with various architectures show that channel and temporal attention modules combined with parallel convolutional pathways yielded the best performance. Beyond its role in learning to interpret ECGs, the findings in this research provide a blueprint for designing interpretable systems in other time-series applications that involve AI.<\/jats:p>","DOI":"10.3389\/frai.2026.1867173","type":"journal-article","created":{"date-parts":[[2026,9,14]],"date-time":"2026-09-14T05:38:57Z","timestamp":1789364337000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["An interpretable multi-scale hybrid attention network (MSHAN) for explainable ECG arrhythmia classification"],"prefix":"10.3389","volume":"9","author":[{"given":"A.","family":"Karmel","sequence":"first","affiliation":[{"name":"School of Computer Science and Engineering, Vellore Institute of Technology","place":["Chennai, India"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Joshita","family":"Das","sequence":"additional","affiliation":[{"name":"School of Computer Science and Engineering, Vellore Institute of Technology","place":["Chennai, India"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Kirtan","family":"Rajesh","sequence":"additional","affiliation":[{"name":"School of Computer Science and Engineering, Vellore Institute of Technology","place":["Chennai, India"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"G.","family":"Kanimozhi","sequence":"additional","affiliation":[{"name":"Centre for Smart Grid Technologies, Vellore Institute of Technology","place":["Chennai, India"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,9,14]]},"reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.36948\/ijfmr.2024.v06i02.18735","article-title":"Explainable artificial intelligence in healthcare","volume":"6","author":"Amit","year":"2024","journal-title":"Int. J. Multidisc. Res."},{"key":"ref2","article-title":"Explainable artificial intelligence (EAI)","volume-title":"Explainable Artificial Intelligence for Biomedical and Healthcare Applications","author":"Amreen","year":"2024"},{"key":"ref3","doi-asserted-by":"publisher","first-page":"208","DOI":"10.1016\/j.hroo.2022.02.004","article-title":"Clinical validation of a novel smartwatch for automated detection of atrial fibrillation","volume":"3","author":"Badertscher","year":"2022","journal-title":"Heart Rhythm."},{"key":"ref4","doi-asserted-by":"publisher","first-page":"103","DOI":"10.1080\/10255842.2024.2378105","article-title":"Arrhythmia detection in inter-patient ECG signals using entropy rate features and RR intervals with CNN architecture","volume":"29","author":"Berrahou","year":"2024","journal-title":"Comput. Methods Biomech. Biomed. Engin."},{"key":"ref5","doi-asserted-by":"publisher","first-page":"764","DOI":"10.3390\/jpm12050764","article-title":"A high-precision deep learning algorithm to localize idiopathic ventricular arrhythmias","volume":"12","author":"Chang","year":"2022","journal-title":"J. Pers. Med."},{"key":"ref9010","doi-asserted-by":"publisher","first-page":"1196","DOI":"10.1109\/TBME.2004.827359","article-title":"Automatic classification of heartbeats using ECG morphology and heartbeat interval features","volume":"51","author":"de Chazal","year":"2004","journal-title":"IEEE Transactions on Biomedical Engineering"},{"key":"ref7","first-page":"1","volume-title":"2024 IEEE International Conference on Contemporary Computing and Communications (InC4)","author":"Fathima","year":"2024"},{"key":"ref8","doi-asserted-by":"crossref","DOI":"10.1109\/AIIoT61789.2024.10578986","volume-title":"AI-Based Detection for Remote Electrocardiogram Monitoring System","author":"Garcia","year":"2024"},{"key":"ref9","first-page":"125062","article-title":"Multi-attention network with redundant information filtering for multi-horizon forecasting in multivariate time series","volume-title":"Expert. Syst. Appl.","author":"Geng","year":"2024"},{"key":"ref10","doi-asserted-by":"publisher","first-page":"754","DOI":"10.1016\/j.ins.2023.01.055","article-title":"Transfer learning based deep network for signal restoration and rhythm analysis during cardiopulmonary resuscitation using only the ECG waveform","volume":"626","author":"Gong","year":"2023","journal-title":"Inf. Sci."},{"key":"ref11","doi-asserted-by":"publisher","first-page":"7233","DOI":"10.3390\/app14167233","article-title":"Detection of arrhythmias using heart rate signals from smartwatches","volume":"14","author":"Herwin","year":"2024","journal-title":"Appl. Sci."},{"key":"ref12","doi-asserted-by":"publisher","first-page":"1387","DOI":"10.1007\/s00530-020-00713-1","article-title":"A novel method for ECG signal classification via one-dimensional convolutional neural network","volume":"28","author":"Hua","year":"2020","journal-title":"Multim. Syst."},{"key":"ref9013","first-page":"3","volume-title":"Lecture Notes in Computer Science","author":"Huang","year":"2019"},{"key":"ref14","doi-asserted-by":"crossref","DOI":"10.1109\/WCONF61366.2024.10691998","article-title":"Deep learning and attention mechanisms in time series modeling","volume-title":"2024 2nd World Conference on Communication and Computing (WCONF)","author":"Jebaraj","year":"2024"},{"key":"ref15","doi-asserted-by":"publisher","first-page":"59842","DOI":"10.1109\/access.2024.3392965","article-title":"ECG classification exercise health analysis algorithm based on GRU and convolutional neural network","volume":"12","author":"Ji","year":"2024","journal-title":"IEEE Access"},{"key":"ref16","first-page":"106745","article-title":"Multi-scale convolution enhanced transformer for multivariate long-term time series forecasting","volume-title":"Neural Netw","author":"Li","year":"2024"},{"key":"ref6","doi-asserted-by":"publisher","first-page":"e143437","DOI":"10.5812\/intjcardiovascpract-143437","article-title":"Advancements in artificial intelligence for ECG signal analysis and arrhythmia detection: a review","volume":"8","author":"Kazemi","year":"2023","journal-title":"Int. J. Cardiovasc. Pract."},{"key":"ref17","first-page":"14","article-title":"Multidimensional time series classification with multiple attention mechanism","volume-title":"Complex Intell. Syst","author":"Liu","year":"2025"},{"key":"ref18","doi-asserted-by":"publisher","first-page":"175","DOI":"10.3390\/jcdd10040175","article-title":"Current and future use of artificial intelligence in electrocardiography","volume":"10","author":"Mart\u00ednez-Sell\u00e9s","year":"2023","journal-title":"J Cardiovasc Dev Dis."},{"key":"ref19","doi-asserted-by":"publisher","first-page":"064006","DOI":"10.1088\/1361-6579\/ac6e55","article-title":"Comparison of neural basis expansion analysis for interpretable time series (N-BEATS) and recurrent neural networks for heart dysfunction classification","volume":"43","author":"Puszkarski","year":"2022","journal-title":"Physiol. Meas."},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1093\/eurheartj\/ehad655.2966","article-title":"AI remote ECG monitoring improves arrhythmia detection","author":"Rosengarten","year":"2023","journal-title":"Eur. Heart J."},{"key":"ref21","doi-asserted-by":"publisher","first-page":"515","DOI":"10.1109\/jbhi.2019.2911367","article-title":"LSTM-based ECG classification for continuous monitoring on personal wearable devices","volume":"24","author":"Saadatnejad","year":"2019","journal-title":"IEEE J. Biomed. Health Inform."},{"key":"ref22","doi-asserted-by":"publisher","first-page":"135","DOI":"10.1007\/s13246-020-00964-2","article-title":"Detection of ventricular arrhythmia using hybrid time\u2013frequency-based features and deep neural network","volume":"44","author":"Sabut","year":"2021","journal-title":"Phys. Eng. Sci. Med."},{"key":"ref23","doi-asserted-by":"publisher","first-page":"385","DOI":"10.1016\/j.jelectrocard.2007.03.008","article-title":"Errors in the computerized electrocardiogram interpretation of cardiac rhythm","volume":"40","author":"Shah","year":"2007","journal-title":"J. Electrocardiol."},{"key":"ref24","doi-asserted-by":"publisher","first-page":"e026974","DOI":"10.1161\/JAHA.122.026974","article-title":"Convolution neural network algorithm for shockable arrhythmia classification within a digitally connected automated external defibrillator","volume":"12","author":"Shen","year":"2023","journal-title":"J. Am. Heart Assoc."},{"key":"ref25","first-page":"203","article-title":"Enhancing patient care and treatment through explainable AI: a gap analysis","volume-title":"Artificial Intelligence in Medicine","author":"Shyni","year":"2024"},{"key":"ref26","doi-asserted-by":"publisher","first-page":"1179","DOI":"10.1093\/europace\/euaa377","article-title":"Deep learning and the electrocardiogram: review of the current state-of-the-art","volume":"23","author":"Somani","year":"2021","journal-title":"Europace"},{"key":"ref27","doi-asserted-by":"publisher","first-page":"25281","DOI":"10.36948\/ijfmr.2024.v06i04.25281","article-title":"Explainable AI: developing interpretable deep learning models for medical diagnosis","volume":"6","author":"Thakur","year":"2024","journal-title":"Int. J. Multidisc. Res."},{"key":"ref28","volume-title":"Advances in Healthcare Information Systems and Administration","author":"Upendran","year":"2024"},{"key":"ref9029","doi-asserted-by":"crossref","first-page":"1288","DOI":"10.23919\/EUSIPCO55093.2022.9909814","volume-title":"2022 30th European Signal Processing Conference (EUSIPCO)","author":"Wang","year":"2022"},{"key":"ref30","first-page":"1341624","article-title":"Multi-submodel fusing attention mechanism for time series prediction","volume-title":"Proc SPIE","author":"Yang","year":"2024"},{"key":"ref31","first-page":"2400527","article-title":"MGGTSP-CAT: integrating temporal convolution and LSTM for multi-scale greenhouse gas time series prediction via cross-attention mechanism","volume-title":"Adv. Theory Simul","author":"Zhong","year":"2024"}],"container-title":["Frontiers in Artificial Intelligence"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1867173\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,14]],"date-time":"2026-09-14T05:39:00Z","timestamp":1789364340000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1867173\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,14]]},"references-count":32,"alternative-id":["10.3389\/frai.2026.1867173"],"URL":"https:\/\/doi.org\/10.3389\/frai.2026.1867173","relation":{},"ISSN":["2624-8212"],"issn-type":[{"value":"2624-8212","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,14]]},"article-number":"1867173"}}