Artificial Intelligence in Cardiology: Clinical Advantages and Perspectives. A Literature Review

Background. The rapid development of artificial intelligence (AI) and machine learning (ML) opens new possibilities in cardiology, enabling the analysis of large volumes of heterogeneous data and the detection of hidden patterns. Unlike traditional statistical methods, which rely on a limited set of factors and linear models, AI can account for hundreds of parameters and nonlinear relationships.Aim. To analyze current approaches to the use of AI in cardiology and demonstrate their clinical advantages compared to traditional statistical methods.Materials and Methods. A systematic literature search was conducted in PubMed, Scopus, Web of Science, and Google Scholar for the period 2015–2025. Keywords included combinations of the terms "artificial intelligence", "machine learning", "cardiology", "ECG", "echocardiography", "CT", "MRI", "prediction". Inclusion criteria: original studies with quantitative assessment of AI model performance and comparison with traditional methods.Results. AI algorithms demonstrate higher accuracy compared to traditional methods: in arrhythmia recognition (AUC ~0.97, F1=0.837 vs 0.780 for physicians), detection of asymptomatic systolic dysfunction (AUC 0.93), and silent atrial fibrillation (AUC 0.87-0.90). In echocardiography, automated systems provide measurement accuracy equivalent to manual analysis but with reduced analysis time to 8 seconds per patient and can detect signs of coronary stenosis (sensitivity 84 %, specificity 93 %, AUC 0.93). In CT coronary angiography, AI achieves specificity of 99.8 % for detecting hemodynamically significant stenoses. Clinical implementation of AI leads to a 15 % reduction in cardiovascular events, 20 % reduction in mortality, and 20 % reduction in hospital readmissions.Conclusions. Artificial intelligence has substantial potential to improve contemporary cardiology practice, particularly in ECG analysis, echocardiography, and coronary CT angiography. Its most promising application is as a clinical decision support tool that complements physicians' expertise and facilitates personalized patient management. Broader implementation of AI requires external validation of predictive models, appropriate regulatory frameworks, and the development of national cardiovascular registries to enable training and validation of AI algorithms on representative patient populations.