A narrative review on the use of artificial intelligence in cardiovascular medicine

Abstract

Background and Objective: Cardiovascular diseases (CVDs) remain the leading cause of morbidity and mortality worldwide despite substantial advancements in prevention, diagnostics, and therapeutics. The integration of artificial intelligence (AI) and machine learning (ML) is transforming cardiovascular medicine, with applications spanning electrocardiogram (ECG) interpretation, advanced imaging analysis, and risk prediction modelling. Authoritative guidelines, observational studies, systematic reviews, and meta-analyses have highlighted the diagnostic and prognostic potential in various domains, including AI-derived physiological age from ECG, automated plaque quantification in coronary computed tomography angiography (CTCA), and time to event survival prediction models. This review aims to synthesise contemporary evidence on AI in cardiovascular diagnostics, prevention, and rehabilitation, and to outline the methodological, ethical, and translational considerations for safe clinical adoption. Methods: A narrative review was conducted. The literature search was performed on PubMed/MEDLINE, Scopus, Embase, and Web of Science using relevant keywords, and articles published between January 2015, and August 2025 were included. Peer-reviewed studies, systematic reviews and meta-analyses, and position statements pertinent to AI in cardiovascular medicine were selected. Key Content and Findings: AI has achieved high accuracy in imaging interpretation, ECG-based arrhythmia detection, multimodal risk stratification, wearable-based screening, and adaptive cardiac rehabilitation. Survival models, such as Random Survival Forests and DeepSurv, have outperformed traditional Cox models in select datasets. Additionally, AI-derived physiological age from ECG shows associations with incident cardiovascular events and mortality. However, external validation is inconsistent, calibration is often inadequate, and reporting standards are variable. Equity, data privacy, interpretability, and workflow integration remain substantial barriers. Conclusions: AI can augment but not replace clinician judgment in cardiovascular care. Translation into routine practice requires rigorous multi-centre prospective trials, transparent reporting checklists, fairness assessments, and governance frameworks to ensure safety, generalizability, and equity. AI will likely play a significantly increasing role in enhancing patients’ cardiovascular care.

Ähnliche Arbeiten