Artificial intelligence in cardiovascular imaging and risk stratification: From algorithmic discovery to meaningful clinical implication

作者：Xia X, Khan WU, Khan Q, Zhong C, Khan MS · 发表于：Trends in cardiovascular medicine · 年份：2026 · DOI：10.1016/j.tcm.2026.08.005 · 研究领域：Artificial intelligence、Cardiac MRI、Cardiovascular imaging、Coronary CT angiography、Echocardiography、Meaningful clinical implication、Risk stratification

Cardiovascular disease (CVD) remains the leading global cause of mortality, with timely diagnosis and precise risk stratification serving as cornerstones of effective management. Traditional cardiovascular imaging and risk assessment have long been constrained by operator-dependent interpretation, time-intensive manual quantification, and a reliance on static, population-derived diagnostic thresholds. The rapid maturation of artificial intelligence (AI), particularly deep learning (DL), foundation models (FMs), and multimodal integration (MI), has catalyzed a paradigm shift toward automated, quantitative, and patient-specific cardiovascular evaluation. Between 2023 and 2026, AI-driven tools have progressed from retrospective proof-of-concept studies to early prospective clinical validations across echocardiography (EchoCG), cardiac magnetic resonance (CMR), coronary computed tomography angiography (CCTA), and nuclear imaging. Concurrently, AI-enhanced electrocardiography (ECG) and polygenic risk integration have enabled dynamic, longitudinal risk prediction models that outperform conventional scores. Despite these advances, clinical adoption faces substantial hurdles including algorithmic bias, limited generalizability across diverse populations, regulatory fragmentation, workflow integration challenges, and unresolved questions regarding clinical utility and cost-effectiveness. This review critically examines the technological evolution of AI in cardiovascular imaging, evalu...