Soft Tissue Radiodensitometric Longitudinal Differences are Associated with Cardiovascular Risks in Elderly Subjects

The increasing prevalence of cardiovascular diseases among aging populations poses significant challenges to healthcare systems. This study investigates the role of soft tissue composition and its longitudinal changes in relation to hypertension (HTN), coronary heart disease (CHD), and cardiovascular disease (CVD). Using the AGES-Reykjavik dataset, we analyzed 3,157 older adults over a five-year period, extracting 11 nonlinear trimodal regression analysis (NTRA) parameters from mid-thigh CT scans to assess muscle, fat, and connective tissue quality. Statistical analysis revealed that muscle degeneration is significant across all cardiovascular conditions, with connective tissue distribution showing distinct patterns associated with the three HTN classes and the binary CHD classes. Machine learning (ML) models demonstrated strong predictive capabilities, with Extra-Trees (EX-T) achieving over 0.75 of accuracy in the binary classification of both the cardiovascular conditions. These findings reinforce the relevance of NTRA-based radiodensity features for cardiovascular risk assessment and highlight their potential for personalized aging-related health monitoring. Future work will refine classification models and explore alternative data balancing techniques to enhance predictive accuracy.