{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,20]],"date-time":"2026-08-20T16:28:46Z","timestamp":1787243326141,"version":"build-2736575974"},"reference-count":40,"publisher":"JMIR Publications Inc.","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["JMIR Med Inform"],"abstract":"Abstract<\/jats:title>\n \n Background<\/jats:title>\n Atrial fibrillation (AF) is a common arrhythmia associated with an increased risk of stroke and heart failure. To improve prevention, recent studies have used deep learning models to identify at-risk individuals early from normal sinus rhythm (NSR). However, studies using mobile electrocardiogram (mECG) in outpatient, real-world settings remain underexplored.<\/jats:p>\n <\/jats:sec>\n \n Objective<\/jats:title>\n The study aimed to develop and evaluate deep learning models using a real-world limb-lead mECG database to predict the short-term occurrence of AF from NSR recordings.<\/jats:p>\n <\/jats:sec>\n \n Methods<\/jats:title>\n mECG data were collected from real-world users of commercially available handheld mECG devices capable of capturing 6 limb leads. AF occurrence was defined as an AF event within a predefined time window (7, 14, or 31 d) from the date of the NSR recording. Transformer-based prediction models were developed for limb-lead and lead I input configurations using a multistage training approach with self-supervised pretraining and domain adaptation, drawing on both open, large-scale clinical 12-lead ECG and proprietary real-world mECG databases. The models were evaluated in an internal real-world cohort and explored in an external cohort as a proof of concept via time-to-event analysis.<\/jats:p>\n <\/jats:sec>\n \n Results<\/jats:title>\n \n Between March 2023 and November 2024, 386,519 mECGs were acquired from 8206 users. There were 18,949, 25,206, and 33,524 AF incidences within the 7-, 14-, and 31-day time windows. The models were pretrained with 787,257 12-lead ECGs and 202,689 mECGs, then fine-tuned to predict AF occurrence using 97,447 labeled mECGs. The limb-lead models achieved areas under the receiver operating characteristic curves (AUROCs) of 0.793, 0.785, and 0.787 for 7-, 14-, and 31-day predictions on the internal cohort, respectively, with a user-level AUROC of 0.702 for the 31-day prediction. These models significantly outperformed the lead I models (\n P<\/jats:italic>\n <.001), supporting the value of multilead configurations. The multistage pretraining was essential, as single-source pretraining yielded lower AUROCs of 0.555 with mECGs only and 0.761 with 12-lead ECGs only for the 31-day prediction. In the subgroup analysis, AUROC values were consistent across age, PR interval, and corrected QT interval, but showed disparities (\n P<\/jats:italic>\n <.001) by sex (0.713 in females vs 0.794 in males) and by QRS duration (0.583 in \u2265120 ms vs 0.796 in <120 ms). In the external cohort (n=144), the 31-day model stratified all 5 new-onset AF events, showing significantly different survival functions between the positively and negatively predicted groups (\n P<\/jats:italic>\n =.03); Cox proportional hazards regression yielded a hazard ratio of 1.49 (95% CI 1.06\u20102.09) per 0.1 increase in model output.\n <\/jats:p>\n <\/jats:sec>\n \n Conclusions<\/jats:title>\n Our findings elucidate the feasibility of deep learning\u2013based AF risk prediction using single-NSR recordings from mobile devices, highlighting the potential for remote AF management in real-world populations. The model output may serve as a risk indicator to support opportunistic AF screening, prompting further clinical evaluation and informing decisions about more intensive monitoring.<\/jats:p>\n <\/jats:sec>","DOI":"10.2196\/87142","type":"journal-article","created":{"date-parts":[[2026,8,20]],"date-time":"2026-08-20T16:14:35Z","timestamp":1787242475000},"page":"e87142-e87142","source":"Crossref","is-referenced-by-count":0,"title":["Prediction of Atrial Fibrillation Occurrence With Handheld Mobile Electrocardiogram: Deep Learning Model Development Using Real-World Data"],"prefix":"10.2196","volume":"14","author":[{"ORCID":"https:\/\/orcid.org\/0009-0005-8907-7008","authenticated-orcid":false,"given":"Minje","family":"Park","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5920-1585","authenticated-orcid":false,"given":"Hyun Jin","family":"Ahn","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0006-1897-2336","authenticated-orcid":false,"given":"Yeongyeon","family":"Na","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3538-3975","authenticated-orcid":false,"given":"Sunghoon","family":"Joo","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0004-3035-6689","authenticated-orcid":false,"given":"Young Ho","family":"Lee","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-0327-5021","authenticated-orcid":false,"given":"Seongwoo","family":"Han","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8352-5069","authenticated-orcid":false,"given":"Myung Soo","family":"Park","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6713-4300","authenticated-orcid":false,"given":"Dae Young","family":"Cheon","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0384-5407","authenticated-orcid":false,"given":"Jeen Hwa","family":"Lee","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9938-3464","authenticated-orcid":false,"given":"Ki Hong","family":"Lee","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1010","published-online":{"date-parts":[[2026,8,20]]},"reference":[{"issue":"7","key":"R1","doi-asserted-by":"publisher","first-page":"1016","DOI":"10.1093\/europace\/euab036","article-title":"Smartphone electrocardiogram for detecting atrial fibrillation after a cerebral ischaemic event: a multicentre randomized controlled trial","volume":"23","author":"Koh","journal-title":"Europace"},{"issue":"10201","key":"R2","doi-asserted-by":"publisher","first-page":"861","DOI":"10.1016\/S0140-6736(19)31721-0","article-title":"An artificial intelligence-enabled ECG algorithm for the identification of patients with atrial fibrillation during sinus rhythm: a retrospective analysis of outcome prediction","volume":"394","author":"Attia","journal-title":"Lancet"},{"issue":"1","key":"R3","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-021-92172-5","article-title":"A new deep learning algorithm of 12-lead electrocardiogram for identifying atrial fibrillation during sinus rhythm","volume":"11","author":"Baek","journal-title":"Sci Rep"},{"issue":"19","key":"R4","doi-asserted-by":"publisher","DOI":"10.1161\/JAHA.123.034154","article-title":"Machine learning algorithm to predict atrial fibrillation using serial 12-lead ECGs based on left atrial remodeling","volume":"13","author":"Choi","journal-title":"J Am Heart Assoc"},{"key":"R5","doi-asserted-by":"publisher","DOI":"10.3389\/fcvm.2022.865852","article-title":"Evaluating the risk of paroxysmal atrial fibrillation in noncardioembolic ischemic stroke using artificial intelligence\u2013enabled ECG algorithm","volume":"9","author":"Han","journal-title":"Front Cardiovasc Med"},{"key":"R6","doi-asserted-by":"publisher","DOI":"10.3389\/fcvm.2023.1168054","article-title":"Clinical applicability of an artificial intelligence prediction algorithm for early prediction of non-persistent atrial fibrillation","volume":"10","author":"Kim","journal-title":"Front Cardiovasc Med"},{"issue":"17","key":"R7","doi-asserted-by":"publisher","DOI":"10.3390\/diagnostics13172849","article-title":"Predicting future incidences of cardiac arrhythmias using discrete heartbeats from normal sinus rhythm ECG signals via deep learning methods","volume":"13","author":"Kim","journal-title":"Diagnostics (Basel)"},{"issue":"10359","key":"R8","doi-asserted-by":"publisher","first-page":"1206","DOI":"10.1016\/S0140-6736(22)01637-3","article-title":"Artificial intelligence\u2013guided screening for atrial fibrillation using electrocardiogram during sinus rhythm: a prospective non-randomised interventional trial","volume":"400","author":"Noseworthy","journal-title":"Lancet"},{"issue":"5","key":"R9","doi-asserted-by":"publisher","first-page":"414","DOI":"10.1016\/j.medj.2024.02.006","article-title":"Artificial intelligence\u2013based risk prediction and mechanism discovery for atrial fibrillation using heart beat-to-beat intervals","volume":"5","author":"Lin","journal-title":"Med"},{"issue":"5","key":"R10","doi-asserted-by":"publisher","DOI":"10.3346\/jkms.2024.39.e56","article-title":"Identification of atrial fibrillation with single-lead mobile ECG during normal sinus rhythm using deep learning","volume":"39","author":"Kim","journal-title":"J Korean Med Sci"},{"issue":"1","key":"R11","doi-asserted-by":"publisher","first-page":"21","DOI":"10.1016\/j.cvdhj.2023.01.002","article-title":"Artificial intelligence\u2013enabled mobile electrocardiograms for event prediction in paroxysmal atrial fibrillation","volume":"4","author":"Raghunath","journal-title":"Cardiovasc Digit Health J"},{"issue":"12","key":"R12","doi-asserted-by":"publisher","first-page":"9052","DOI":"10.1109\/TPAMI.2024.3415112","article-title":"A survey on self-supervised learning: algorithms, applications, and future trends","volume":"46","author":"Gui","journal-title":"IEEE Trans Pattern Anal Mach Intell"},{"key":"R13","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.01553","article-title":"Masked autoencoders are scalable vision learners","author":"He","journal-title":"IEEE\/CVF Conf Comput Vis Pattern Recognit"},{"key":"R14","unstructured":"Na Y Park M Tae Y Joo S . Guiding masked representation learning to capture spatio-temporal relationship of electrocardiogram. Presented at: The Twelfth International Conference on Learning Representations; May 7-11, 2024; Vienna, Austria. URL: https:\/\/proceedings.iclr.cc\/paper_files\/paper\/2024\/file\/412fb8623bf8b6d56fb6285ea295447e-Paper-Conference.pdf [Accessed 29-07-2026]"},{"key":"R15","doi-asserted-by":"publisher","DOI":"10.1109\/TIM.2022.3228267","article-title":"MaeFE: masked autoencoders family of electrocardiogram for self-supervised pretraining and transfer learning","volume":"72","author":"Zhang","journal-title":"IEEE Trans Instrum Meas"},{"key":"R16","doi-asserted-by":"crossref","unstructured":"Gururangan S Marasovi\u0107 A Swayamdipta S . Don\u2019t stop pretraining: adapt language models to domains and tasks. In: Jurafsky D Chai J Schluter N Tetreault J , editors. Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. 2020:8342-8360. [doi: 10.18653\/v1\/2020.acl-main.740]","DOI":"10.18653\/v1\/2020.acl-main.740"},{"key":"R17","unstructured":"Lee S Kang M Lee J Hwang SJ Kawaguchi K . Self-distillation for further pre-training of transformers. Presented at: 11th International Conference on Learning Representations, ICLR 2023; May 1-5, 2023; Kigali, Rwanda. URL: https:\/\/openreview.net\/pdf?id=kj6oK_Hj40 [Accessed 29-07-2026]"},{"key":"R18","unstructured":"Gow B Pollard T Nathanson LA Johnson A Moody B Fernandes C et al. MIMIC-IV-ECG: diagnostic electrocardiogram matched subset (version 1.0). PhysioNet. URL: https:\/\/physionet.org\/content\/mimic-iv-ecg\/1.0\/ [Accessed 29-07-2026]"},{"issue":"3","key":"R19","doi-asserted-by":"publisher","first-page":"510","DOI":"10.1016\/j.chest.2018.09.011","article-title":"A simple clinical risk score (CHEST) for predicting incident atrial fibrillation in Asian subjects: derivation in 471,446 Chinese subjects, with internal validation and external application in 451,199 Korean subjects","volume":"155","author":"Li","journal-title":"Chest"},{"issue":"1","key":"R20","doi-asserted-by":"publisher","DOI":"10.1097\/MD.0000000000005597","article-title":"Usefulness of HATCH score in the prediction of new-onset atrial fibrillation for Asians","volume":"96","author":"Suenari","journal-title":"Medicine (Baltimore)"},{"issue":"6","key":"R21","doi-asserted-by":"publisher","first-page":"84","DOI":"10.1145\/3065386","article-title":"ImageNet classification with deep convolutional neural networks","volume":"60","author":"Krizhevsky","journal-title":"Commun ACM"},{"key":"R22","unstructured":"Dosovitskiy A Beyer L Kolesnikov A Weissenborn D Zhai X Unterthiner T et al. An image is worth 16x16 words: transformers for image recognition at scale. Presented at: The Ninth International Conference on Learning Representations; May 3-7, 2021. URL: https:\/\/openreview.net\/pdf?id=YicbFdNTTy [Accessed 29-07-2026]"},{"key":"R23","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.00949","article-title":"Class-balanced loss based on effective number of samples","author":"Cui","journal-title":"IEEE\/CVF Conf Comput Vis Pattern Recognit"},{"key":"R24","doi-asserted-by":"publisher","DOI":"10.1109\/CVPRW50498.2020.00359","article-title":"Randaugment: practical automated data augmentation with a reduced search space","author":"Cubuk","journal-title":"IEEE\/CVF Conf Comput Vis Pattern Recognit Workshops"},{"key":"R25","doi-asserted-by":"crossref","unstructured":"Nonaka N Seita J . RandECG: data augmentation for deep neural network based ECG classification. In: Takama Y , editor. Advances in Artificial IntelligenceSelected Papers from the Annual Conference of Japanese Society of Artificial Intelligence (JSAI 2021). 2021:178-189. [doi: 10.1007\/978-3-030-96451-1_16]","DOI":"10.1007\/978-3-030-96451-1_16"},{"key":"R26","unstructured":"Loshchilov I Hutter F . Decoupled weight decay regularization. Presented at: The Seventh International Conference on Learning Representations; May 6-9, 2019; New Orleans, LA, USA. URL: https:\/\/openreview.net\/pdf\/5963886abef941684ffc0cf670297e47fb1e5155.pdf [Accessed 29-07-2026]"},{"key":"R27","unstructured":"Loshchilov I Hutter F . SGDR: Stochastic gradient descent with warm restarts. Presented at: The Fifth International Conference on Learning Representations; Apr 24-26, 2017; Toulon, France. URL: https:\/\/openreview.net\/pdf?id=Skq89Scxx [Accessed 29-07-2026]"},{"key":"R28","unstructured":"Bao H Dong L Piao S Wei F . BEiT: BERT pre-training of image transformers. arXiv. Preprint posted online on Jun 15, 2021. [doi: 10.48550\/arXiv.2106.08254]"},{"issue":"3","key":"R29","doi-asserted-by":"crossref","first-page":"837","DOI":"10.2307\/2531595","volume":"44","author":"DeLong","journal-title":"Biometrics"},{"issue":"4","key":"R30","doi-asserted-by":"publisher","first-page":"404","DOI":"10.1093\/biomet\/26.4.404","article-title":"The use of confidence or fiducial limits illustrated in the case of the binomial","volume":"26","author":"Clopper","journal-title":"Biometrika"},{"issue":"13","key":"R31","doi-asserted-by":"publisher","first-page":"1020","DOI":"10.1136\/heartjnl-2021-320036","article-title":"Prediction of incident atrial fibrillation in community-based electronic health records: a systematic review with meta-analysis","volume":"108","author":"Nadarajah","journal-title":"Heart"},{"issue":"11","key":"R32","doi-asserted-by":"publisher","DOI":"10.1111\/eci.14293","article-title":"Usefulness of the CHEST score to predict new onset atrial fibrillation. A systematic review and meta-analysis on >11 million subjects","volume":"54","author":"Pastori","journal-title":"Eur J Clin Invest"},{"key":"R33","doi-asserted-by":"publisher","DOI":"10.3389\/fcvm.2023.1160242","article-title":"Six-lead electrocardiography compared to single-lead electrocardiography and photoplethysmography of a wrist-worn device for atrial fibrillation detection controlled by premature atrial or ventricular contractions: six is smarter than one","volume":"10","author":"Bacevicius","journal-title":"Front Cardiovasc Med"},{"issue":"4","key":"R34","doi-asserted-by":"publisher","first-page":"643","DOI":"10.1093\/ehjdh\/ztab083","article-title":"Clinical validation and evaluation of a novel six-lead handheld electrocardiogram recorder compared to the 12-lead electrocardiogram in unselected cardiology patients (EVALECG Cardio)","volume":"2","author":"Azram","journal-title":"Eur Heart J Digit Health"},{"key":"R35","doi-asserted-by":"publisher","DOI":"10.3389\/fphys.2025.1532284","article-title":"Reconstruction of 12-lead ECG: a review of algorithms","volume":"16","author":"Obianom","journal-title":"Front Physiol"},{"issue":"4","key":"R36","doi-asserted-by":"publisher","first-page":"621","DOI":"10.1093\/jamia\/ocz228","article-title":"A tutorial on calibration measurements and calibration models for clinical prediction models","volume":"27","author":"Huang","journal-title":"J Am Med Inform Assoc"},{"issue":"10","key":"R37","doi-asserted-by":"publisher","first-page":"1120","DOI":"10.1001\/jamacardio.2017.3180","article-title":"Incidence of previously undiagnosed atrial fibrillation using insertable cardiac monitors in a high-risk population: the REVEAL AF study","volume":"2","author":"Reiffel","journal-title":"JAMA Cardiol"},{"key":"R38","doi-asserted-by":"publisher","DOI":"10.1016\/j.ahj.2018.11.003","article-title":"New atrial fibrillation diagnosed by 30-day rhythm monitoring","volume":"209","author":"Farris","journal-title":"Am Heart J"},{"issue":"7","key":"R39","doi-asserted-by":"publisher","first-page":"100668","DOI":"10.1016\/j.medj.2025.100668","article-title":"Clinician-artificial intelligence collaboration: a win-win solution for efficiency and reliability in atrial fibrillation diagnosis","volume":"6","author":"Zhang","journal-title":"Med"},{"key":"R40","unstructured":"Park M . MAE-based atrial fibrillation prediction. Zenodo. 2026. URL: https:\/\/zenodo.org\/records\/20637293 [Accessed 29-07-2026]"}],"container-title":["JMIR Medical Informatics"],"original-title":[],"language":"en","deposited":{"date-parts":[[2026,8,20]],"date-time":"2026-08-20T16:14:57Z","timestamp":1787242497000},"score":1,"resource":{"primary":{"URL":"https:\/\/medinform.jmir.org\/2026\/1\/e87142"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,20]]},"references-count":40,"URL":"https:\/\/doi.org\/10.2196\/87142","relation":{},"ISSN":["2291-9694"],"issn-type":[{"value":"2291-9694","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,20]]},"article-number":"v14i7e87142"}}