A machine learning-derived aging index for risk stratification and mortality prediction in cardiovascular-kidney-metabolic syndrome: A retrospective cohort study

Researchers have developed a machine learning-based aging index called CKMAI to help predict mortality and the risk of cardiovascular-kidney-metabolic syndrome (CKM) among adults. The study, conducted on a large sample of 6,896 individuals from the National Health and Nutrition Examination Survey (NHANES) between 2005 and 2018, showed that CKMAI outperformed other existing biological age indices in predicting mortality and high-risk CKM status.

Using a two-stage machine learning framework, the researchers evaluated over 100 candidate survival models and selected the optimal model via Pareto front optimization. The CKMAI consistently achieved a mean C-index of 0.893 across all outcomes, including all-cause mortality, cardiovascular mortality, and high-risk CKM status. The index also demonstrated significant time-dependent AUCs and superior net benefit in net reclassification improvement (NRI) and integrated discrimination improvement (IDI) analyses.

The study also found nonlinear associations between CKMAI and the outcomes, with inflection points at CKMAI values of 61.587 for all-cause mortality, 60.732 for cardiovascular mortality, and 34.092 for high-risk CKM status. These thresholds indicated steeper risk increases below each point. Furthermore, CKMAI exhibited super-additive interactions with the PhenoAge index, suggesting a combined effect on mortality risk.

In addition, the research identified six distinct aging-metabolic phenotypes through clustering analysis, with the frail elderly cluster showing the highest mortality risk (hazard ratio of 11.74). The findings also revealed that depression partially mediated the CKMAI-outcome associations, contributing to 5.6% of all-cause mortality risk, 8.2% of cardiovascular mortality risk, and 11.1% of high-risk CKM status risk.

Overall, the CKMAI appears to be a valid and CKM-specific aging index that outperforms universal biological age measures in predicting mortality and identifying high-risk individuals. Its practical implementation as a tool for risk stratification and targeted prevention in CKM syndrome is promising, although more validation in independent cohorts with complete mortality follow-up is needed.