A Structured Review of Machine Learning Methods for Early Heart Disease Prediction

This paper has presented a structured review of machine learning and deep learning methods for early heart disease detection, covering recent primary studies and foundational works across six research directions. The reviewed works indicate that multimodal deep learning approaches, which fuse ECG temporal signals with structured clinical and lifestyle data, can provide promising performance compared with single-modality and classical machine learning approaches on selected cardiovascular benchmarks. Key architectural patterns identified in the literature include 1D-CNN-GRU temporal encoders, Squeeze-and-Excitation FCN designs for structured tabular data, cross-attention fusion for inter-modal dependency modelling, and SMOTE-ENN resampling for class imbalance remediation. SHAP-based explainability has emerged as an essential component for clinical adoption, consistently identifying es-tablished cardiovascular risk factors such as age, cholesterol level, and systolic blood pressure as dominant predictors, thereby validating the clinical coherence of the reviewed models. Despite strong benchmark performance, critical gaps remain in modality-robustness at inference time, external validation across diverse populations, and regulatory and ethical frameworks for real-world deployment. The taxonomy and cross-paper synthesis presented in this survey provide a structured foundation for future work, and the identified gaps point to clear and impactful research directions in federated learning, imaging modality integration, and genomic biomarker incorporation for personalised cardiac risk stratification.

[1]                          P. Wagner, N. Strodthoff, R. Bousseljot, D. Kreiseler, F. I. Lunze, W. Samek, and T. Schaeffter, “PTB-XL, a Large Publicly Available Electrocardiography Dataset,” Scientific Data, vol. 7, no. 154, 2020. [2]                                              G. B. Moody and R. G. Mark, “The MIT-BIH Arrhythmia Database on CD-ROM and Software for Use with It,” Computers in Cardiology, vol. 17, pp. 185–188, 1990. [3]                                              T. R. Dawber, G. F. Meadors, and F. E. Moore, “Epidemiological Approaches to Heart Disease: The Framingham Study,” American Journal of Public Health, vol. 41, no. 3, pp. 279–286, 1951. [4]                          World Health Organization, “Cardiovascular Diseases (CVDs) Fact Sheet,” WHO, 2022. [Online]. Available: https://www.who.int/news-room/ fact-sheets/detail/cardiovascular-diseases-(cvds) [5]                                              M. B. Barla and A. K. Madam, “A Hybrid Deep Learning Framework for Early Cardiovascular Disease Prediction: SDCN-CHDMO-ResVGG Approach,” in Proc. 9th Int. Conf. Electronics, Communication and Aerospace Technology (ICECA), IEEE, 2025, pp. 1049–1053. [6]                                              Y. Wu, M. Huang, Y. Shi, Q. Li, Y. Zhang, and B. Song, “An Interpretable Hybrid Deep Learning Method for Heart Disease Prediction,” in Proc. 5th Int. Conf. Digital Society and Intelligent Systems (DSInS), IEEE, 2025, pp. 248–252. [7]                                              Ch. Vasavi and G. Sujatha, “Data Driven Heart Disease Prediction: A Comparative Study of ML Models,” in Proc. 4th Int. Conf. Innovative Mechanisms for Industry Applications (ICIMIA), IEEE, 2025, pp. 1711–1715. [8]                          I. Sherly S., T. A. Mohanaprakash, T. Sindhu, N. Subramanian, D. Bora, and K. Cinthuja, “Data Transformer – AI Driven Multimodal Fusion Network for Cardiovascular Risk Prediction Using Multimodal Data,” in Proc. 5th Int. Conf. Evolutionary Computing and Mobile Sustainable Networks (ICECMSN), IEEE, 2025, pp. 203–212. [9]                          L. Karthikeyan and E. Sathiyamoorthy, “Deep Learning for Predicting Cardiovascular Events in Patients with Neurological Disorders: A Multimodal Analysis of Brain-Heart Interactions,” in 2025 Innovations in Power and Advanced Computing Technologies (i-PACT), IEEE, 2025, pp. 1–5. [10]                                           C. Thirumarai Selvi, P. Seleena, R. S. Sankara Subramaninan, A. K. Shruthika, P. Gandhi Ayyavu, and S. Srinidhi, “Heart Disease Prediction using Feature-Optimized ML and Hybrid Deep Learning Models,” in Proc. 3rd Int. Conf. Sustainable Computing and Data Communication Systems (ICSCDS), IEEE, 2025, pp. 1025–1028.