{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,11]],"date-time":"2026-09-11T16:17:46Z","timestamp":1789143466297,"version":"build-2803163510"},"reference-count":31,"publisher":"Public Library of Science (PLoS)","issue":"2","license":[{"start":{"date-parts":[[2026,2,25]],"date-time":"2026-02-25T00:00:00Z","timestamp":1771977600000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100000274","name":"British Heart Foundation","doi-asserted-by":"publisher","award":["FS\/CRTF\/23\/24482"],"award-info":[{"award-number":["FS\/CRTF\/23\/24482"]}],"id":[{"id":"10.13039\/501100000274","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100006041","name":"Innovate UK","doi-asserted-by":"publisher","award":["Reference 10082834"],"award-info":[{"award-number":["Reference 10082834"]}],"id":[{"id":"10.13039\/501100006041","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["www.plosdigitalhealth.org"],"crossmark-restriction":false},"short-container-title":["PLOS Digit Health"],"abstract":"<jats:p>Deep neural networks can classify ECGs with high accuracy when training data is abundant. Rare conditions like Brugada syndrome, an inherited arrhythmia syndrome predisposing to sudden death, pose challenges due to data scarcity hindering model training. We evaluated multiple machine learning (ML) approaches to optimise a Brugada ECG classification model using limited training data. The baseline model was trained on a dataset comprising 176 Brugada, 176 right bundle branch block (RBBB) and 352 normal ECGs from Zhongshan Hospital (Zhongshan-baseline dataset), framed as a binary classification task to distinguish Brugada from non-Brugada ECGs. A 25%-75% train-test split was used to exacerbate data scarcity. To enhance training, we incorporated three additional datasets: (i) a different, labelled ECG dataset from Zhongshan Hospital including normal and RBBB ECGs (Zhongshan-pretrain), (ii) an unlabelled ECG dataset from Hammersmith Hospital including Brugada and non-Brugada ECGs (Imperial), (iii) an open-access labelled ECG dataset (PTB-XL). Three strategies were tested: (1) supervised pretraining, (2) self-supervised pretraining with data augmentation, and (3) oversampling using SMOTE (synthetic minority oversampling technique). Each model was evaluated on the unseen internal test set and an external Brugada mimic dataset. The models were re-trained using an 80%-20% train-test split as a secondary analysis. The baseline model achieved 92.2% accuracy, F1-score 0.837, and area under the Receiver Operating Characteristic curve (AUC) 0.962. Supervised pretraining significantly improved performance when training data was scarce, with the best model pretrained on the Zhongshan-pretrain dataset boosting accuracy (+3.2%), F1-score (+0.071) and AUC\u2009+\u20090.019), with consistent cross-validation performance. Self-supervised pretraining produced smaller and more variable gains, although select models better mitigated against false positives on the Brugada mimic dataset. SMOTE oversampling showed inconsistent effects on performance. Incorporating pretraining and oversampling may facilitate the development of more accurate AI-ECG models for rare diseases when training data is limited but provides diminishing returns when adequate labelled data is available.<\/jats:p>","DOI":"10.1371\/journal.pdig.0001222","type":"journal-article","created":{"date-parts":[[2026,2,25]],"date-time":"2026-02-25T18:34:26Z","timestamp":1772044466000},"page":"e0001222","update-policy":"https:\/\/doi.org\/10.1371\/journal.pdig.corrections_policy","source":"Crossref","is-referenced-by-count":2,"title":["AI-ECG classification for Brugada syndrome: A study of machine learning techniques to optimise for limited datasets"],"prefix":"10.1371","volume":"5","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-9385-9276","authenticated-orcid":true,"given":"Keenan","family":"Saleh","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Raaif","family":"Hadadi","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yixiu","family":"Liang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Hong","family":"Wong","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Arunashis","family":"Sau","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"James","family":"Howard","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Evan","family":"Brittain","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jeffrey","family":"Annis","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Majd","family":"El-Harasis","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Matthew","family":"Shun-Shin","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jagdeep","family":"Mohal","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Akriti","family":"Naraen","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9090-9610","authenticated-orcid":true,"given":"Jack","family":"Samways","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jessica","family":"Artico","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"James","family":"Ware","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Prapa","family":"Kanagaratnam","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Fu Siong","family":"Ng","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0904-2904","authenticated-orcid":true,"given":"Massoud","family":"Zolgharni","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Wenjia","family":"Bai","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Amanda","family":"Varnava","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zachary","family":"Whinnett","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7597-0170","authenticated-orcid":true,"given":"Ahran","family":"Arnold","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"340","published-online":{"date-parts":[[2026,2,25]]},"reference":[{"issue":"46","key":"pdig.0001222.ref001","doi-asserted-by":"crossref","first-page":"4717","DOI":"10.1093\/eurheartj\/ehab649","article-title":"Application of artificial intelligence to the electrocardiogram","volume":"42","author":"ZI Attia","year":"2021","journal-title":"Eur Heart J"},{"key":"pdig.0001222.ref002","doi-asserted-by":"crossref","first-page":"192","DOI":"10.1109\/RBME.2024.3492381","article-title":"A Survey of Few-Shot Learning for Biomedical Time Series","volume":"18","author":"C Li","year":"2025","journal-title":"IEEE Rev Biomed Eng"},{"key":"pdig.0001222.ref003","doi-asserted-by":"crossref","first-page":"2892","DOI":"10.1016\/j.csbj.2024.07.005","article-title":"Synthetic data generation methods in healthcare: A review on open-source tools and methods","volume":"23","author":"VC Pezoulas","year":"2024","journal-title":"Comput Struct Biotechnol J"},{"issue":"7","key":"pdig.0001222.ref004","doi-asserted-by":"crossref","first-page":"925","DOI":"10.1093\/europace\/eus042","article-title":"About Brugada syndrome and its prevalence","volume":"14","author":"PG Postema","year":"2012","journal-title":"Europace"},{"key":"pdig.0001222.ref005","doi-asserted-by":"crossref","first-page":"1010748","DOI":"10.3389\/fcvm.2022.1010748","article-title":"Sex-related differences in incidence, phenotype and risk of sudden cardiac death in inherited arrhythmia syndromes","volume":"9","author":"B Asatryan","year":"2023","journal-title":"Front Cardiovasc Med"},{"issue":"10","key":"pdig.0001222.ref006","doi-asserted-by":"crossref","first-page":"1468","DOI":"10.1016\/j.hrthm.2019.07.003","article-title":"Ethnic differences in patients with Brugada syndrome and arrhythmic events: New insights from Survey on Arrhythmic Events in Brugada Syndrome","volume":"16","author":"A Milman","year":"2019","journal-title":"Heart Rhythm"},{"issue":"1","key":"pdig.0001222.ref007","doi-asserted-by":"crossref","first-page":"154","DOI":"10.1038\/s41597-020-0495-6","article-title":"PTB-XL, a large publicly available electrocardiography dataset","volume":"7","author":"P Wagner","year":"2020","journal-title":"Sci Data"},{"issue":"1","key":"pdig.0001222.ref008","article-title":"An Open-source Toolbox for Analysing and Processing PhysioNet Databases in MATLAB and Octave","volume":"2","author":"I Silva","year":"2014","journal-title":"J Open Res Softw"},{"key":"pdig.0001222.ref009","doi-asserted-by":"crossref","unstructured":"Liu CM, Liu CL, Hu KW, Tseng VS, Chang SL, Lin YJ, et al. A Deep Learning\u2013Enabled Electrocardiogram Model for the Identification of a Rare Inherited Arrhythmia: Brugada Syndrome. Canadian Journal of Cardiology. 2022;38 (2): 152\u20139. https:\/\/doi.org\/10.1016\/J.CJCA.2021.08.014","DOI":"10.1016\/j.cjca.2021.08.014"},{"key":"pdig.0001222.ref010","first-page":"1575","author":"T Chen","year":"2020","journal-title":"A simple framework for contrastive learning of visual representations"},{"key":"pdig.0001222.ref011","author":"X Chen","year":"2020"},{"key":"pdig.0001222.ref012","doi-asserted-by":"crossref","first-page":"103010","DOI":"10.1016\/j.bspc.2021.103010","article-title":"Self-supervised ECG pre-training","volume":"70","author":"H Liu","year":"2021","journal-title":"Biomedical Signal Processing and Control"},{"issue":"2","key":"pdig.0001222.ref013","doi-asserted-by":"crossref","first-page":"373","DOI":"10.1109\/TAI.2022.3159505","article-title":"Detection of Cardiovascular Diseases in ECG Images Using Machine Learning and Deep Learning Methods","volume":"4","author":"MB Abubaker","year":"2023","journal-title":"IEEE Trans Artif Intell"},{"key":"pdig.0001222.ref014","doi-asserted-by":"crossref","first-page":"105114","DOI":"10.1016\/j.compbiomed.2021.105114","article-title":"Self-supervised representation learning from 12-lead ECG data","volume":"141","author":"T Mehari","year":"2022","journal-title":"Comput Biol Med"},{"issue":"4","key":"pdig.0001222.ref015","doi-asserted-by":"crossref","first-page":"1129","DOI":"10.1007\/s13246-019-00815-9","article-title":"Automatic detection of arrhythmia from imbalanced ECG database using CNN model with SMOTE","volume":"42","author":"SK Pandey","year":"2019","journal-title":"Australas Phys Eng Sci Med"},{"key":"pdig.0001222.ref016","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1109\/TIM.2022.3181276","article-title":"Automatic Cardiac Arrhythmia Classification Using Residual Network Combined With Long Short-Term Memory","volume":"71","author":"YK Kim","year":"2022","journal-title":"IEEE Trans Instrum Meas"},{"key":"pdig.0001222.ref017","first-page":"9525","author":"J Adebayo","year":"2018","journal-title":"Proceedings of the 32nd International Conference on Neural Information Processing Systems"},{"issue":"4","key":"pdig.0001222.ref018","doi-asserted-by":"crossref","first-page":"404","DOI":"10.1016\/j.ihj.2012.06.010","article-title":"Brugada pattern masquerading as ST-segment elevation myocardial infarction in flecainide toxicity","volume":"64","author":"L Chhabra","year":"2012","journal-title":"Indian Heart J"},{"issue":"3","key":"pdig.0001222.ref019","first-page":"775","article-title":"Brugada syndrome: manifest, concealed, \u201casymptomatic,\u201d suspected and simulated","volume":"38","author":"B Surawicz","year":"2001","journal-title":"J Am Coll Cardiol"},{"issue":"7","key":"pdig.0001222.ref020","article-title":"Brugada Syndrome Misdiagnosed As Acute Myocardial Infarction: A Case Report","volume":"14","author":"AR Khan","year":"2022","journal-title":"Cureus"},{"issue":"7","key":"pdig.0001222.ref021","doi-asserted-by":"crossref","first-page":"1095","DOI":"10.1093\/europace\/euv278","article-title":"Expert cardiologists cannot distinguish between Brugada phenocopy and Brugada syndrome electrocardiogram patterns","volume":"18","author":"BH Gottschalk","year":"2016","journal-title":"Europace"},{"issue":"2","key":"pdig.0001222.ref022","doi-asserted-by":"crossref","first-page":"100410","DOI":"10.1016\/j.patter.2021.100410","article-title":"Breaking away from labels: The promise of self-supervised machine learning in intelligent health","volume":"3","author":"D Spathis","year":"2022","journal-title":"Patterns (N Y)"},{"key":"pdig.0001222.ref023","author":"HAAK Hammoud","year":"2024","journal-title":"On Pretraining Data Diversity for Self-Supervised Learning"},{"key":"pdig.0001222.ref024","author":"W Morningstar","year":"2024"},{"issue":"11","key":"pdig.0001222.ref025","doi-asserted-by":"crossref","first-page":"453","DOI":"10.3390\/bios11110453","article-title":"Automatic Multi-Label ECG Classification with Category Imbalance and Cost-Sensitive Thresholding","volume":"11","author":"Y Liu","year":"2021","journal-title":"Biosensors (Basel)"},{"issue":"4","key":"pdig.0001222.ref026","doi-asserted-by":"crossref","first-page":"683","DOI":"10.1093\/ehjdh\/ztaf039","article-title":"Synthetic electrocardiograms for Brugada syndrome: from data generation to expert cardiologists evaluation","volume":"6","author":"B Zanchi","year":"2025","journal-title":"Eur Heart J Digit Health"},{"key":"pdig.0001222.ref027","doi-asserted-by":"crossref","first-page":"109453","DOI":"10.1016\/j.compbiomed.2024.109453","article-title":"Synthetic ECG signals generation: A scoping review","volume":"184","author":"B Zanchi","year":"2025","journal-title":"Comput Biol Med"},{"key":"pdig.0001222.ref028","doi-asserted-by":"crossref","unstructured":"Liao S, Bokhari M, Chakraborty P, Suszko A, Jones G, Spears D, et al. Use of Wearable Technology and Deep Learning to Improve the Diagnosis of Brugada Syndrome. JACC: Clinical Electrophysiology. 2022;8 (8): 1010\u201320. https:\/\/doi.org\/10.1016\/J.JACEP.2022.05.003","DOI":"10.1016\/j.jacep.2022.05.003"},{"key":"pdig.0001222.ref029","doi-asserted-by":"crossref","unstructured":"Ronan R, Tarabanis C, Chinitz L, Jankelson L, Charney LH. Brugada ECG detection with self-supervised VICReg pre-training: a novel deep learning approach for rare cardiac diseases. medRxiv. 2024; 2024.03.29.24305072. https:\/\/doi.org\/10.1101\/2024.03.29.24305072","DOI":"10.1101\/2024.03.29.24305072"},{"key":"pdig.0001222.ref030","doi-asserted-by":"crossref","first-page":"223","DOI":"10.1007\/978-90-481-9485-8_14","article-title":"Rare diseases social epidemiology: analysis of inequalities","volume":"686","author":"A Kole","year":"2010","journal-title":"Adv Exp Med Biol"},{"issue":"4","key":"pdig.0001222.ref031","doi-asserted-by":"crossref","first-page":"619","DOI":"10.1093\/ehjdh\/ztaf051","article-title":"A deep foundation model for electrocardiogram interpretation: enabling rare disease detection through transfer learning","volume":"6","author":"SM Hu","year":"2025","journal-title":"Eur Heart J Digit Health"}],"container-title":["PLOS Digital Health"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pdig.0001222","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,2,25]],"date-time":"2026-02-25T18:34:35Z","timestamp":1772044475000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pdig.0001222"}},"subtitle":[],"editor":[{"given":"Philipp","family":"Berens","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2026,2,25]]},"references-count":31,"journal-issue":{"issue":"2","published-online":{"date-parts":[[2026,2,25]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pdig.0001222","relation":{},"ISSN":["2767-3170"],"issn-type":[{"value":"2767-3170","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,2,25]]}}}