A scoping review on aging and cardiovascular diseases - Molecular mediators and artificial intelligence-based advanced diagnostic methods

Background: Cardiovascular diseases (CVDs) remain the leading cause of mortality worldwide, and their risk increases with age. Biological age reflects physiological decline more accurately than chronological age and may improve cardiovascular risk prediction. Recent advances in Artificial Intelligence (AI) have enabled estimation of biological age from electrocardiograms (ECGs), imaging, biomarkers, and omics data. However, the existing evidence on AI-derived biological age in cardiovascular medicine has not been comprehensively synthesized. Methods: PubMed and Scopus were searched for studies published between 2019 and 2025. Eligible studies included original research applying AI or machine learning to estimate biological age or aging biomarkers in human participants and relating these estimates to cardiovascular outcomes. Data were extracted on study design, AI methods, input data sources, biological age metrics, cardiovascular endpoints, and validation strategies. Findings were synthesized narratively and organized by methodological approaches. Results: AI-derived biological age, particularly ECG-predicted age, consistently provided prognostic information beyond chronological age. Deep learning models identified an age gap associated with increased risks of heart failure, atrial fibrillation and stroke. Retinal imaging-based biological age and biomarker-based machine learning models further supported the systemic nature of aging. Emerging approaches such as generative models offer insights into aging trajectories and lifestyle-linked aging phenotypes. Advances in interpretability using explainable AI highlighted ECG features that contribute most to aging predictions. Conclusions: AI-based estimation of biological age is a promising biomarker for cardiovascular risk assessment. Although evidence across modalities is consistent, further standardization, prospective validation, and improved interpretability are needed to support clinical implementation.