POSSIBILITIES OF APPLYING MACHINE LEARNING FOR ELECTROCARDIOGRAM ANALYSIS TO PREDICT MYOCARDIAL INFARCTION

Abstract

Introduction. Cardiovascular disease, including acute myocardial infarction, remains the leading cause of death worldwide, and early and accurate recognition of ischemic changes on electrocardiography is critical to the effectiveness of emergency care. Conventional visual interpretation of electrocardiograms is limited by subjectivity and reduced sensitivity to subtle or atypical patterns, prompting the development of novel analytical approaches based on artificial intelligence. Aim. To summarize current evidence on the potential of machine learning methods for electrocardiogram signal analysis in the prediction and early diagnosis of myocardial infarction, provide specific examples of clinical applications of artificial intelligence in cardiology, and identify key unresolved issues in this field. Materials and methods. A literature search was performed in PubMed, Scopus, Web of Science, and Google Scholar, supplemented by an analysis of official data from the World Health Organization, the Centers for Disease Control and Prevention, and the American Heart Association published between 2021 and 2026. The search used the keywords “machine learning,” “deep learning,” “electrocardiogram,” “myocardial infarction,” and “artificial intelligence.” A total of 46 relevant sources were included in this narrative review. Results. Convolutional neural network-, ensemble-, and transformer-based models have demonstrated the ability to detect electrocardiographic features of ischemia and myocardial infarction, as well as other cardiac abnormalities, including left ventricular dysfunction, myocardial hypertrophy, and atrial fibrillation. In several studies, their diagnostic performance exceeded that of conventional expert interpretation. The first pragmatic randomized clinical trial demonstrated that an artificial intelligence–based electrocardiogram alert intervention was associated with reduced all-cause mortality. Key unresolved issues include limited generalizability of models to new patient populations, a shortage of prospective validation studies, insufficient algorithmic interpretability, and the lack of standardized approaches to clinical implementation. Conclusions. Machine learning represents a promising tool for improving the accuracy of electrocardiogram-based prediction and early detection of myocardial infarction. However, broad clinical implementation requires methodological standardization, further prospective validation, and the development of interpretable and clinically applicable algorithms.