基于人工智能的生成式心电数据增强方法研究综述

doi:10.3778/j.issn.1002-8331.2501-0188

摘要/Abstract

摘要： 基于心电图的心血管疾病智能诊断依赖于高质量数据，但患者隐私保护、高质量心电数据稀缺及类别不平衡是其面临的主要挑战，心电数据增强被广泛用于解决这些问题。介绍了常用的数据库和质量评价指标；回顾了生成对抗网络、数学拟合模型、心脏电生理模型、变分自编码器和扩散模型五种生成式心电数据增强方法；对比分析了不同方法对应的评估指标与实验结果、数据库、生成心电图导联数和模型输入，以及不同模型的优缺点和应用场景，结果表明生成对抗网络是最常用的心电数据增强模型，改进的扩散模型正成为心电数据增强的研究热点，心脏电生理模型适合用于提升可解释性；展望了未来研究方向，包括生成ECG生理真实性与临床相关性的加强，模型稳定性与多样性的协同提升，轻量化与边缘计算的定向适配，可解释生成和临床决策的结合。

关键词: 心电图, 数据增强, 生成对抗网络, 心脏电生理模型, 扩散模型

Abstract: Intelligent diagnosis for cardiovascular disease based on electrocardiogram (ECG) relies on the high-quality data, but patient privacy protection, scarcity of high-quality ECG data, and class imbalance pose major challenges. ECG data augmentation has been widely employed to address these issues. This paper introduces commonly used databases and quality evaluation metrics, reviews five generative ECG augmentation methods, including generative adversarial networks (GANs), mathematical fitting models, cardiac electrophysiological models, variational autoencoders, and diffusion models. The paper provides a comparative analysis of the evaluation metrics, experimental results, databases, number of generated ECG leads, and model inputs corresponding to different methods. The advantages, disadvantages, and application scenarios of different models are also discussed. The results indicate that GANs are the most commonly used models for ECG data augmentation, improved diffusion models are becoming a research hotspot in ECG data augmentation, and cardiac electrophysiological models are suitable for enhancing interpretability. Finally, future research directions are highlighted, including enhancing the physiological realism and clinical relevance of synthetic ECGs, achieving the synergistic improvement in model stability and diversity, enabling the targeted adaptation of lightweight architectures for edge computing, and integrating interpretable generation with clinical decision-making.

Key words: electrocardiogram, data augmentation, generative adversarial network, cardiac electrophysiology model, diffusion model

韩闯, 付瑞雪, 周钰森, 阙文戈. 基于人工智能的生成式心电数据增强方法研究综述[J]. 计算机工程与应用, 2025, 61(16): 64-75.

HAN Chuang, FU Ruixue, ZHOU Yusen, QUE Wenge. Review of Generative Electrocardiogram Data Augmentation Methods Based on Artificial Intelligence[J]. Computer Engineering and Applications, 2025, 61(16): 64-75.

参考文献

[1] BENJAMIN E J, MUNTNER P, ALONSO A, et al. Heart disease and stroke statistics-2019 update: a report from the American heart association[J]. Circulation, 2019, 139(10): e56-e528.
[2] HANNUN A Y, RAJPURKAR P, HAGHPANAHI M, et al. Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network[J]. Nature Medicine, 2019, 25(1): 65-69.
[3] ACHARYA U R, FUJITA H, OH S L, et al. Application of deep convolutional neural network for automated detection of myocardial infarction using ECG signals[J]. Information Sciences, 2017, 415: 190-198.
[4] SHIN H C, TENENHOLTZ N A, ROGERS J K, et al. Medical image synthesis for data augmentation and anonymization using generative adversarial networks[C]//Proceedings of the 3rd International Workshop on Simulation and Synthesis in Medical Imaging. Cham: Springer, 2018: 1-11.
[5] WANG G J, CHEN M, DING Z J, et al. Inter-patient ECG arrhythmia heartbeat classification based on unsupervised domain adaptation[J]. Neurocomputing, 2021, 454: 339-349.
[6] YU J, GAO J, WANG N, et al. Spa-Tem MI: a spatial-temporal network for detecting and locating myocardial infarction[J]. IEEE Transactions on Instrumentation and Measurement, 2023, 72: 2522712.
[7] SARKAR P, ETEMAD A. Self-supervised ECG representation learning for emotion recognition[J]. IEEE Transactions on Affective Computing, 2022, 13(3): 1541-1554.
[8] ZHANG W R, GENG S J, HONG S D. A simple self-supervised ECG representation learning method via manipulated temporal-spatial reverse detection[J]. Biomedical Signal Processing and Control, 2023, 79: 104194.
[9] RAHMAN M M, RIVOLTA M W, BADILINI F, et al. A systematic survey of data augmentation of ECG signals for AI applications[J]. Sensors, 2023, 23(11): 5237.
[10] BERGER L, HABERBUSCH M, MOSCATO F. Generative adversarial networks in electrocardiogram synthesis: recent developments and challenges[J]. Artificial Intelligence in Medicine, 2023, 143: 102632.
[11] MONACHINO G, ZANCHI B, FIORILLO L, et al. Deep generative models: the winning key for large and easily accessible ECG datasets?[J]. Computers in Biology and Medicine, 2023, 167: 107655.
[12] 陈景霞, 唐喆喆, 林文涛, 等. 用于脑电数据增强和情绪识别的自注意力GAN[J]. 计算机工程与应用, 2023, 59(5): 160-168.
CHEN J X, TANG Z Z, LIN W T, et al. Self-attention GAN for EEG data augmentation and emotion recognition[J]. Computer Engineering and Applications, 2023, 59(5): 160-168.
[13] 陈鹏, 邓淼磊, 樊好义, 等. 心电特征引导下的自监督房颤异常检测方法[J]. 计算机工程与应用, 2025, 61(2): 208-218.
CHEN P, DENG M L, FAN H Y, et al. Self-supervised atrial fibrillation anomaly detection method guided by electrocardiogram features[J]. Computer Engineering and Applications, 2025, 61(2): 208-218.
[14] MOODY G B, MARK R G. The impact of the MIT-BIH arrhythmia database[J]. IEEE Engineering in Medicine and Biology Magazine, 2001, 20(3): 45-50.
[15] BOUSSELJOT R, KREISELER D, SCHNABEL A. Nutzung der EKG-signaldatenbank cardiodat der PTB über das Internet[J]. Biomedizinische Technik, 1995, 40: 317-318.
[16] WAGNER P, STRODTHOFF N, BOUSSELJOT R D, et al. PTB-XL, a large publicly available electrocardiography dataset[J]. Scientific Data, 2020, 7: 154.
[17] CLIFFORD G D, LIU C Y, MOODY B, et al. AF classification from a short single lead ECG recording: the PhysioNet/computing in cardiology challenge 2017[J]. Computing in Cardiology, 2017. DOI:10.22489/CinC.2017.065-469.
[18] GOLDBERGER A L, AMARAL L A, GLASS L, et al. PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals[J]. Circulation, 2000, 101(23): E215-E220.
[19] LIU F F, LIU C Y, ZHAO L N, et al. An open access database for evaluating the algorithms of electrocardiogram rhythm and morphology abnormality detection[J]. Journal of Medical Imaging and Health Informatics, 2018, 8(7): 1368-1373.
[20] ALT H, GODAU M. Computing the Fréchet distance between two polygonal curves[J]. International Journal of Computational Geometry & Applications, 1995, 5: 75-91.
[21] MüLLER M. Dynamic time warping[M]//Information retrieval for music and motion. Berlin, Heidelberg: Springer, 2007: 69-84.
[22] KIM M G, PAN S B. A study on user recognition using the generated synthetic electrocardiogram signal[J]. Sensors, 2021, 21(5): 1887.
[23] GOODFELLOW I, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial nets[C]//Advances in Neural Information Processing Systems 27, 2014.
[24] XIA X, PAN X Z, LI N, et al. GAN-based anomaly detection: a review[J]. Neurocomputing, 2022, 493: 497-535.
[25] MIRZA M, OSINDERO S. Conditional generative adversarial nets[J]. arXiv:1411.1784, 2014.
[26] HOSSAIN K F, KAMRAN S A, TAVAKKOLI A, et al. ECG-adv-GAN: detecting ECG adversarial examples with conditional generative adversarial networks[C]//Proceedings of the 2021 20th IEEE International Conference on Machine Learning and Applications. Piscataway: IEEE, 2021: 50-56.
[27] ZHOU X, ZHU X, NAKAMURA K, et al. Electrocardiogram quality assessment with a generalized deep learning model assisted by conditional generative adversarial networks[J]. Life, 2021, 11(10): 1013.
[28] WANG X Y, CHEN B C, ZENG M, et al. An ECG signal denoising method using conditional generative adversarial net[J]. IEEE Journal of Biomedical and Health Informatics, 2022, 26(7): 2929-2940.
[29] WU J, WANG L P, PAN H L, et al. MLCGAN: multi-lead ECG synthesis with multi label conditional generative adversarial network[C]//Proceedings of the 2023 IEEE International Conference on Acoustics, Speech and Signal Processing. Piscataway: IEEE, 2023: 1-5.
[30] ZHAN Z H, CHEN J R, LI K M, et al. Conditional generative adversarial network driven variable-duration single-lead to 12-lead electrocardiogram reconstruction[J]. Biomedical Signal Processing and Control, 2024, 95: 106377.
[31] ODENA A, OLAH C, SHLENS J. Conditional image synthesis with auxiliary classifier GANs[C]//Proceedings of the 34th International Conference on Machine Learning, 2017: 2642-2651.
[32] FRID-ADAR M, DIAMANT I, KLANG E, et al. GAN-based synthetic medical image augmentation for increased CNN performance in liver lesion classification[J]. Neurocomputing, 2018, 321: 321-331.
[33] WANG W, SHENG Y Q, WANG J L, et al. HAST-IDS: learning hierarchical spatial-temporal features using deep neural networks to improve intrusion detection[J]. IEEE Access, 2017, 6: 1792-1806.
[34] DONG H W, HSIAO W Y, YANG L C, et al. MuseGAN: symbolic-domain music generation and accompaniment with multi-track sequential generative adversarial networks[J]. arXiv:1709.06298, 2017.
[35] WANG P, HOU B R, SHAO S Y, et al. ECG arrhythmias detection using auxiliary classifier generative adversarial network and residual network[J]. IEEE Access, 2019, 7: 100910-100922.
[36] ZHOU Z H, ZHAI X L, TIN C. Fully automatic electrocardiogram classification system based on generative adversarial network with auxiliary classifier[J]. Expert Systems with Applications, 2021, 174: 114809.
[37] RADFORD A, METZ L, CHINTALA S, et al. Unsupervised representation learning with deep convolutional generative adversarial networks[J]. arXiv:1511.06434, 2015.
[38] WULAN N R, WANG W, SUN P Z, et al. Generating electrocardiogram signals by deep learning[J]. Neurocomputing, 2020, 404: 122-136.
[39] DASGUPTA S, DAS S, BHATTACHARYA U. CardioGAN: an attention-based generative adversarial network for generation of electrocardiograms[C]//Proceedings of the 2020 25th International Conference on Pattern Recognition. Piscataway: IEEE, 2021: 3193-3200.
[40] THAMBAWITA V, ISAKSEN J L, HICKS S A, et al. DeepFake electrocardiograms using generative adversarial networks are the beginning of the end for privacy issues in medicine[J]. Scientific Reports, 2021, 11: 21896.
[41] LI W Q, TANG Y M, YU K M, et al. SLC-GAN: an automated myocardial infarction detection model based on generative adversarial networks and convolutional neural networks with single-lead electrocardiogram synthesis[J]. Information Sciences, 2022, 589: 738-750.
[42] MA S, CUI J F, XIAO W D, et al. Deep learning-based data augmentation and model fusion for automatic arrhythmia identification and classification algorithms[J]. Computational Intelligence and Neuroscience, 2022(1): 1577778.
[43] 杨春德, 贾竹, 李欣蔚. 基于U-Net++的心电信号识别分类研究[J]. 计算机科学, 2021, 48(10): 121-126.
YANG C D, JIA Z, LI X W. Study on ECG signal recognition and classification based on U-Net++[J]. Computer Science, 2021, 48(10): 121-126.
[44] QIN J, GAO F J, WANG Z M, et al. A novel temporal generative adversarial network for electrocardiography anomaly detection[J]. Artificial Intelligence in Medicine, 2023, 136: 102489.
[45] BASAK P, SAKIB A N, CHOWDHURY M E H, et al. A novel deep learning technique for morphology preserved fetal ECG extraction from mother ECG using 1D-CycleGAN[J]. Expert Systems with Applications, 2024, 235: 121196.
[46] GOLANY T, RADINSKY K, FREEDMAN D. SimGANs: simulator-based generative adversarial networks for ECG synthesis to improve deep ECG classification[C]//Proceedings of the 37th International Conference on Machine Learning, 2020: 3597-3606.
[47] GOLANY T, FREEDMAN D, RADINSKY K. ECG ODE-GAN: learning ordinary differential equations of ECG dynamics via generative adversarial learning[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2021, 35(1): 134-141.
[48] KARRAS T, AILA T M, LAINE S, et al. Progressive growing of GANs for improved quality, stability, and variation[J]. arXiv:1710.10196, 2017.
[49] YANG H X, LIU J H, ZHANG L H, et al. ProEGAN-MS: a progressive growing generative adversarial networks for electrocardiogram generation[J]. IEEE Access, 2021, 9: 52089-52100.
[50] QU Z G, LI Y, LIU B, et al. DTQFL: a digital twin-assisted quantum federated learning algorithm for intelligent diagnosis in 5G mobile network[J]. IEEE Journal of Biomedical and Health Informatics, 2023. DOI:10.1109/JBHI.2023.3303401.
[51] QU Z G, LI Y, TIWARI P. QNMF: a quantum neural network based multimodal fusion system for intelligent diagnosis[J]. Information Fusion, 2023, 100: 101913.
[52] LLOYD S, WEEDBROOK C. Quantum generative adversarial learning[J]. Physical Review Letters, 2018, 121(4): 040502.
[53] QU Z G, SHI W K, TIWARI P. Quantum conditional generative adversarial network based on patch method for abnormal electrocardiogram generation[J]. Computers in Biology and Medicine, 2023, 166: 107549.
[54] QU Z G, CHEN W L, TIWARI P. HQ-DCGAN: hybrid quantum deep convolutional generative adversarial network approach for ECG generation[J]. Knowledge-Based Systems, 2024, 301: 112260.
[55] MCSHARRY P E, CLIFFORD G D, TARASSENKO L, et al. A dynamical model for generating synthetic electrocardiogram signals[J]. IEEE Transactions on Bio-Medical Engineering, 2003, 50(3): 289-294.
[56] 马佳越. 基于时序自动机心脏模型的心律失常识别[D]. 秦皇岛: 燕山大学, 2021.
MA J Y. Arrhythmia recognition based on timed automata heart model[D]. Qinhuangdao: Yanshan University, 2021.
[57] 徐永红, 王金萍, 马佳越. 基于时序心脏模型样本均衡方法的心律失常分类[J]. 中国生物医学工程学报, 2022, 41(3): 301-309.
XU Y H, WANG J P, MA J Y. Arrhythmia classification based on time-series cardiac model sample equalization[J]. Chinese Journal of Biomedical Engineering, 2022, 41(3): 301-309.
[58] LOPEZ-PEREZ A, SEBASTIAN R, IZQUIERDO M, et al. Personalized cardiac computational models: from clinical data to simulation of infarct-related ventricular tachycardia[J]. Frontiers in Physiology, 2019, 10: 580.
[59] D?SSEL O, LUONGO G, NAGEL C, et al. Computer modeling of the heart for ECG interpretation: a review[J]. Hearts, 2021, 2(3): 350-368.
[60] LOEWE A, WüLFERS E M, SEEMANN G. Cardiac ischemia: insights from computational models[J]. Herzschr Elektrophys, 2018, 29(1): 48-56.
[61] QUE W G, HAN C, ZHAO X L, et al. An ECG generative model of myocardial infarction[J]. Computer Methods and Programs in Biomedicine, 2022, 225: 107062.
[62] QUE W G, BIAN Y N, CHEN S J, et al. Efficient electrocardiogram generation based on cardiac electric vector simulation model[J]. Computers in Biology and Medicine, 2024, 177: 108629.
[63] KUZNETSOV V V, MOSKALENKO V A, ZOLOTYKH N Y. Electrocardiogram generation and feature extraction using a variational autoencoder[J]. arXiv:2002.00254, 2020.
[64] LIU H, ZHAO Z B, CHEN X, et al. Using the VQ-VAE to improve the recognition of abnormalities in short-duration 12-lead electrocardiogram records[J]. Computer Methods and Programs in Biomedicine, 2020, 196: 105639.
[65] XIA Y, WANG W Y, WANG K Q. ECG signal generation based on conditional generative models[J]. Biomedical Signal Processing and Control, 2023, 82: 104587.
[66] AL NAZI Z, BISWAS A, ABU RAYHAN M, et al. Classification of ECG signals by dot residual LSTM network with data augmentation for anomaly detection[C]//Proceedings of the 2019 22nd International Conference on Computer and Information Technology. Piscataway: IEEE, 2019: 1-5.
[67] ALCARAZ J M L, STRODTHOFF N. Diffusion-based conditional ECG generation with structured state space models[J]. Computers in Biology and Medicine, 2023, 163: 107115.
[68] NEIFAR N, BEN-HAMADOU A, MDHAFFAR A, et al. DiffECG: a versatile probabilistic diffusion model for ECG signals synthesis[J]. arXiv:2306.01875, 2023.
[69] ADIB E, FERNANDEZ A S, AFGHAH F, et al. Synthetic ECG signal generation using probabilistic diffusion models[J]. IEEE Access, 2023, 11: 75818-75828.
[70] ZAMA M H, SCHWENKER F. ECG synthesis via diffusion-based state space augmented transformer[J]. Sensors, 2023, 23(19): 8328.
[71] SEO H C, YOON G W, JOO S, et al. Multiple electrocardiogram generator with single-lead electrocardiogram[J]. Computer Methods and Programs in Biomedicine, 2022, 221: 106858.
[72] ZHOU F Y, LI J J. ECG data enhancement method using generate adversarial networks based on Bi-LSTM and CBAM[J]. Physiological Measurement, 2024, 45(2): 025003.