AI-Powered Analysis of Metabolic Risk Factors in Non-CVD Individuals and Their Trajectory Toward Cardiovascular Incidence
Code: G-1179
Authors: Maryam Mahdavi ℗, Anoshirvan Kazemnejad *, Abbas Asosheh, Davood Khalili
Schedule: Not Scheduled!
Tag: Biomedical Signal Processing
Download: Download Poster
Abstract:
Abstract
Background and aims: Cardiovascular disease (CVD) is the leading global cause of mortality, responsible for 18.6 million deaths in 2019, with increasing prevalence and incidence. In Iran, CVD accounts for 46.04% of all deaths, with demographic aging and sedentary lifestyles exacerbating the burden. This study evaluates the impact of metabolic risk factors and their trajectories on CVD development in an Iranian cohort. Methods: Based on the Tehran Lipid and Glucose Study (TLGS), this longitudinal study included 1872 adults aged 40–79 years without prior CVD at baseline. Participants were selected through multistage random cluster sampling and followed from 1999 to 2018. Data were collected on demographic, lifestyle, and metabolic factors, with laboratory analyses conducted using standardized protocols. Generalized Estimating Equations (GEE) were used to assess age- and gender-adjusted trajectories of metabolic indicators, while Random Survival Forest (RSF) models evaluated the predictive performance of CVD risk factors. Harrell’s C-index and residual analysis compared the full and reduced RSF models. Results: During a 10-year follow-up, 117 participants (6.3%) developed CVD. Baseline CVD converters exhibited higher age, weight, blood pressure, fasting glucose, lipid levels, and diabetes prevalence. Key metabolic risk factor trajectories included TyG Index, FPG, and SBP, which had significant increases 6 years before diagnosis. RSF model performance was robust, with C-index values of 0.95 (full model) and 0.92 (reduced model), and a Pearson correlation of 0.80 between model predictions. Residual analysis showed slight variability but strong overall alignment between models. Conclusion: Longitudinal trajectories of metabolic risk factors, particularly SBP, FPG, and TyG index, demonstrated strong predictive value for CVD development years before onset, with SBP emerging as the most potent predictor. These findings emphasize the importance of early detection and preventive strategies targeting metabolic risk factors. Lifestyle modifications can significantly mitigate CVD risk, underscoring the utility of longitudinal data in understanding risk factor heterogeneity and disease progression.
Keywords
Trajectory, Cardiovascular-disease, TLGS, Risk-Factor, Artificial Intelligence
