Early Diagnosis of Congenital Heart Disease Using Transformer-Based Deep Learning on Electrocardiogram Signals

Abstract

Congenital heart disease(CHD)is one of the most common birth defects worldwide and a major cause of pediatric morbidity and mortality. Early detection is essential for improving survival and long-term outcomes, yet timely diagnosis remains challenging, especially in primary-care and resource-constrained settings. Although electrocardiography(ECG)is inexpensive and widely available, subtle pediatric CHD abnormalities are difficult to detect through manual interpretation, which carries subjectivity and risk of misdiagnosis. Moreover, many existing deep-learning models rely on single-domain convolutional neural network(CNN)or recurrent neural network(RNN)architectures that insufficiently capture long-range temporal dependencies and frequency-domain features. To address these limitations, we propose pediatric AI for cardiac ECG recognition(PACER), a hybrid CNNTransformer-discrete wavelet transform(DWT)-TabNet framework for automated CHD detection from standard 12-lead pediatric ECG signals. PACER integrates convolutional layers for local morphological extraction, Transformerbased self-attention for long-range temporal modeling, DWT for frequency representation, and TabNet for interpretable multimodal feature fusion. A tailored preprocessing and augmentation pipeline, including SMOTE and Gaussian noise enhancement, improves robustness to class imbalance. Evaluated on 10 344 pediatric ECG recordings using stratified cross-validation, PACER achieved an accuracy of 90. 93%, an F1 score of 0. 91, and an area under the receiver operating characteristic curve(ROC-AUC)of 0. 95, outperforming CNN, RNN, hybrid, Transformer, and CHDdECG baselines. Ablation experiments and model interpretability analysis validate the effectiveness of each module and indicate the potential clinical utility of the proposed method.

Keywords

congenital heart disease / deep learning / Transformer networks / pediatric electrocardiography / automated diagnosis

References

[1] Liu Y J, Chen S, Zühlke L, et al. Global birth prevalence of congenital heart defects 1970 - 2017: updated systematic review and meta-analysis of 260 studies[J]. International Journal of Epidemiology, 2019, 48(2): 455-463.

[2] Lv H S, Sun F Y, Chen Y. Decoding congenital heart disease: a multi-omic framework for cardiac lineage and regulatory dysfunction[J]. Frontiers in Cell and Developmental Biology, 2025, 13: 1659884.

[3] Chen J T, Huang S, Zhang Y, et al. Congenital heart disease detection by pediatric electrocardiogram based deep learning integrated with human concepts[J]. Nature Communications, 2024, 15: 976.

[4] Song L H, Wang Y, Wang H, et al. Clinical profile of congenital heart diseases detected in a tertiary hospital in China: a retrospective analysis[J]. Frontiers in Cardiovascular Medicine, 2023, 10: 1131383.

[5] Aly S, Qattea I, Kattea M O, et al. Neonatal outcomes in preterm infants with severe congenital heart disease: a national cohort analysis[J]. Frontiers in Pediatrics, 2024, 12: 1326804.

[6] Li F, Wang S, Gao Z, et al. Harnessing artificial intelligence in sepsis care: advances in early detection, personalized treatment, and realtime monitoring[J]. Frontiers in Medicine, 2025, 11: 1510792.

[7] Oster M E, Pinto N M, Pramanik A K, et al. Newborn Screening for critical congenital heart disease: a new algorithm and other updated recommendations: clinical report[J]. Pediatrics, 2025, 155: e2024069667.

[8] Zhang Y Y, Wang J Y, Zhao J X, et al. Current status and challenges in prenatal and neonatal screening, diagnosis, and management of congenital heart disease in China[J]. The Lancet Child & Adolescent Health, 2023, 7(7): 479-489.

[9] Khan K, Ullah F, Syed I, et al. Accurately assessing congenital heart disease using artificial intelligence[J]. PeerJ Computer Science, 2024, 10: e2535.

[10] Rao P S. Advances in the diagnosis and management of congenital heart disease in children[J]. Children, 2023, 10(4): 753.

[11] Sachdeva R, Armstrong A K, Arnaout R, et al. Novel Techniques in Imaging Congenital Heart Disease[J]. Journal of the American College of Cardiology, 2024, 83(1): 63-81.

[12] Negussie M, Sanchez N, Sidiq S A, et al. Applying artificial intelligence to cardiac MRI to diagnose congenital heart disease in low-resource settings such as Sub-Saharan Africa[J]. Communications Medicine, 2025, 5: 473.

[13] Holzer R J, Bergersen L, Thomson J, et al. PICS/AEPC/APPCS/CSANZ/SCAI/SOLACI: expert consensus statement on cardiac catheterization for pediatric patients and adults with congenital heart disease[J]. JACC: Cardiovascular Interventions, 2024, 17(2): 115-216.

[14] Marnani R A, Jaros R, Pavlicek J, et al. Advancements and challenges in non-invasive electrocardiography for prenatal, intrapartum, and postnatal care: a comprehensive review[J]. IEEE Access, 2024, 12: 44730-44747.

[15] Alkahtani H K, Haq I U, Ghadi Y Y, et al. Precision diagnosis: an automated method for detecting congenital heart diseases in children from phonocardiogram signals employing deep neural network[J]. IEEE Access, 2024, 12: 76053-76064.

[16] Pachiyannan P, Alsulami M, Alsadie D, et al. A novel machine learning-based prediction method for early detection and diagnosis of congenital heart disease using ECG signal processing[J]. Technologies, 2024, 12(1): 4.

[17] Leone D M, O’Sullivan D, Bravo-Jaimes K. Artificial intelligence in pediatric electrocardiography: a comprehensive review[J]. Children, 2024, 12(1): 25.

[18] Pachiyannan P, Alsulami M, Alsadie D, et al. A cardiac deep learning model(CDLM)to predict and identify the risk factor of congenital heart disease[J]. Diagnostics, 2023, 13(13): 2195.

[19] Karna V V R, Karna V R, Janamala V, et al. A comprehensive review on heart disease risk prediction using machine learning and deep learning algorithms[J]. Archives of Computational Methods in Engineering, 2025, 32(3): 1763-1795.

[20] Song Y S, Liu G H. DCA-YOLO: detection algorithm for YOLOv8 pulmonary nodules based on attention mechanism optimization[J]. Journal of Donghua University(English Edition), 2025, 42(1): 78-87.

[21] Rahman T, Al-Ruweidi M K A A, Sumon Md S I, et al. Deep learning technique for congenital heart disease detection using stacking-based CNNLSTM models from fetal echocardiogram: a pilot study[J]. IEEE Access, 2023, 11: 110375-110390.

[22] Ameen A, Fattoh I E, Abd El-Hafeez T, et al. Advances in ECG and PCG-based cardiovascular disease classification: a review of deep learning and machine learning methods[J]. Journal of Big Data, 2024, 11(1): 159.

[23] Haq I U, Husnain G, Ghadi Y Y, et al. Enhancing pediatric congenital heart disease detection using customized 1D CNN algorithm and phonocardiogram signals[J]. Heliyon, 2025, 11(3): e42257.

[24] Lee Y S, Chung H T, Lin J J, et al. Prediction of significant congenital heart disease in infants and children using continuous wavelet transform and deep convolutional neural network with 12-lead electrocardiogram[J]. BMC Pediatrics, 2025, 25(1): 324.

[25] Alomar K, Aysel H I, Cai X H. CNNs, RNNs and Transformers in human action recognition: a survey and a hybrid model[J]. Artificial Intelligence Review, 2025, 58(12): 387.

[26] Malekifar M, Cao S, Phan K, et al. A survey of deep learning for non-invasive fetal electrocardiogram analysis[J]. IEEE Access, 2025, 13: 202553-202575.

[27] Vaswani A, Shazeer N, Parmar N, et al.Attention is all you need[PP/OL]. arXiv(2017-06-12)[2026-01-13].

[28] Gao Y, Yang H, Pan J, et al. Diagnosis and classification of congenital heart disease in infants based on Gated Swin-Transformer with Multiscale Feature Fusion[J]. Biomedical Signal Processing and Control, 2025, 108: 107919.

[29] Alkhodari M, Hadjileontiadis L J, Khandoker A H. Identification of congenital valvular murmurs in young patients using deep learning-based attention transformers and phonocardiograms[J]. IEEE Journal of Biomedical and Health Informatics, 2024, 28(4): 1803-1814.

[30] Ansari M Y, Yaqoob M, Ishaq M, et al. A survey of transformers and large language models for ECG diagnosis: advances, challenges, and future directions[J]. Artificial Intelligence Review, 2025, 58(9): 261.

[31] Oporto E, Mauricio D, Maculan N, et al. Challenges in the classification of cardiac arrhythmias and ischemia using end-to-end deep learning and the electrocardiogram: a systematic review[J]. Diagnostics, 2026, 16(1): 161.

[32] Han Y, Murino V, Liu X, et al. A systematic review on foundation models for electrocardiogram analysis: initial strides and expansive horizons[PP/OL]. arXiv,(2024-10-24)[2026-01-13].

[33] Ejaz H, Thyyib T, Ibrahim A, et al. Role of artificial intelligence in early detection of congenital heart diseases in neonates[J]. Frontiers in Digital Health, 2024, 5: 1345814.

[34] Jabbar A, Grooby E, Poh Y Y, et al. Automated detection of pediatric congenital heart disease from phonocardiograms using deep and handcrafted feature fusion[J]. Computers in Biology and Medicine, 2025, 197: 110993.

[35] Zhang S T, Kang C X, Cui J, et al. Development of machine learning-based models to predict congenital heart disease: a matched case-control study[J]. International Journal of Medical Informatics, 2025, 195: 105741.

[36] Lam C, Gurvitz M. Transition from pediatric to adult cardiology care in congenital heart disease: contemporary considerations[J]. European Journal of Pediatrics, 2025, 184(7): 466.

[37] Antoun I, Nizam A, Ebeid A, et al. Artificial intelligence in adult congenital heart disease: diagnostic and therapeutic applications and future directions[J]. Reviews in Cardiovascular Medicine, 2025, 26(8): 41523.

[38] Sunilkumar G, Kumaresan P. Deep learning and transfer learning in cardiology: a review of cardiovascular disease prediction models[J]. IEEE Access, 2024, 12: 193365-193386.

[39] Mondal S, Maity R, Nag A. A comprehensive survey of heart disease prediction approaches: methods, applications, performance analysis, datasets, research challenges, and future scopes[J/OL]. Archives of Computational Methods in Engineering, 2025-11-08.

[40] Oikonomou E K, Khera R, Expanding artificial intelligence to understudied populations: congenital heart disease as the next frontier[J]. European Heart Journal, 2025, 46(9): 869-871.

[41] Khalid M, Pluempitiwiriyawej C, Wangsiripitak S, et al. The applications of deep learning in ecg classification for disease diagnosis: a systematic review and meta-data analysis[J]. Engineering Journal, 2024, 28(8): 45-77.

[42] Kalimuthu M, Hemanth C. A comparative analysis of machine learning and deep learning approaches for phonocardiogram classification using dataset integration[J]. IEEE Access, 2025, 13: 170619-170635.

[43] Sarafraz G, Behnamnia A, Hosseinzadeh M, et al. Domain adaptation and generalization of functional medical data: a systematic survey of brain data[J]. ACM Computing Surveys, 2024, 56(10): 1-39.

[44] Kim J, Kim H, Kim H, et al. A comprehensive survey of deep learning for time series forecasting: architectural diversity and open challenges[J]. Artificial Intelligence Review, 2025, 58(7): 216.

[45] Cao K J, Zhang T, Huang J. Advanced hybrid LSTM-transformer architecture for real-time multi-task prediction in engineering systems[J]. Scientific Reports, 2024, 14(1): 4890.

[46] Patro B N, Namboodiri V P, Agneeswaran V S. SpectFormer: frequency and attention is what you need in a vision transformer[C]//2025 IEEE/CVF Winter Conference on Applications of Computer Vision(WACV). Piscataway, NJ, USA: IEEE, 2025: 9543-9554.

[47] Yemets K, Izonin I, Dronyuk I. Time series forecasting model based on the adapted transformer neural network and FFT-based features extraction[J]. Sensors, 2025, 25(3): 652.

[48] Cheng M, Wang J, Liu X, Et Al. Development and validation of a deep-learning network for detecting congenital heart disease from multi-view multi-modal transthoracic echocardiograms[J]. Research, 2024, 7: 319.

[49] Mayourian J, Asztalos I B, El-Bokl A, et al. Electrocardiogram-based deep learning to predict left ventricular systolic dysfunction in paediatric and adult congenital heart disease in the USA: a multicentre modelling study[J]. Lancet Digital Health, 2025, 7(4): e264-e274.

[50] Arivalagan D, Ochathevan V, Dhanasekaran R. Identification of cardiac risk factors from ECG signals using residual neural networks[J]. Congenital Heart Disease, 2025, 20(4): 477-501.

[51] Ghelani H. Automated defect detection in printed circuit boards: exploring the impact of convolutional neural networks on quality assurance and environmental sustainability in manufacturing[J]. International Journal of Advanced Engineering Technologies and Innovations, 2022, 1(4): 275-289.