Machine-learning CKMAI aging index predicts mortality in CKM syndrome cohort
In a weighted NHANES 2005–2018 cohort of 6,896 adults, a machine-learning aging index (CKMAI) tailored to cardiovascular-kidney-metabolic syndrome predicted all-cause and cardiovascular mortality and high-risk CKM status better than PhenoAge, KDM and CMI. [1]
Automatic summary · not editorially reviewed
The key findings
The retrospective cohort included 6,896 adults from NHANES 2005–2018, weighted for the US adult population, and used a two-stage machine-learning framework to build the CKMAI. [1]
CKMAI consistently outperformed PhenoAge, KDM and the cardiometabolic index across all outcomes, with time-dependent AUCs of 0.893 at 3 years, 0.907 at 5 years and 0.890 at 10 years for all-cause mortality. [1]
Clustering revealed six aging-metabolic phenotypes, and the frail elderly cluster had the highest mortality risk (hazard ratio 11.74, 95% CI 7.85–17.57). [1]
The source at a glance
- Study type / format
- Retrospective cohort study (abstract).
- Population / context
- 6,896 adults from NHANES 2005–2018, weighted for the US adult population.
- Summary based on
- Study abstract
- Original publication
- October 6, 2026
Retrospective NHANES cohort of 6,896 adults
This was a retrospective cohort study of 6,896 adults from NHANES 2005–2018, weighted for the US adult population. The authors built a CKM-specific aging index (CKMAI) with a two-stage machine-learning framework that evaluated over 100 candidate survival models; the optimal model was chosen by Pareto front optimization. Predictive performance was assessed with time-dependent AUC, net reclassification improvement, integrated discrimination improvement, calibration and decision curve analysis. The abstract reports a mean C-index of 0.893 for the Pareto-optimal simple Cox model.
Supporting source [1]
CKMAI outperformed PhenoAge, KDM and CMI
CKMAI consistently outperformed PhenoAge, KDM and the cardiometabolic index across all outcomes. For all-cause mortality, time-dependent AUCs were 0.893 (95% CI 0.851–0.925) at 3 years, 0.907 (0.884–0.927) at 5 years and 0.890 (0.868–0.910) at 10 years. For cardiovascular mortality they were 0.904 (0.840–0.943), 0.937 (0.892–0.960) and 0.937 (0.900–0.959); for high-risk CKM status, 0.790 (0.765–0.816). Reclassification and discrimination improvements were significant at all time points, and decision curve analysis showed superior net benefit.
Supporting source [1]
Thresholds, phenotypes and partial depression mediation
Nonlinear associations were found for all three outcomes, with steeper risk increases below inflection points at CKMAI 61.587 (all-cause mortality), 60.732 (cardiovascular mortality) and 34.092 (high-risk CKM status). CKMAI showed a super-additive interaction with PhenoAge (relative excess risk due to interaction 15.06, 95% CI 8.14–38.44). Clustering revealed six aging-metabolic phenotypes; the frail elderly cluster had the highest mortality risk (hazard ratio 11.74, 95% CI 7.85–17.57). Depression partially mediated the associations, with mediated proportions of 5.6% for all-cause mortality, 8.2% for cardiovascular mortality and 11.1% for high-risk CKM status.
Supporting source [1]
How to interpret this
Abstract only; the full text was not reviewed. A key stated limitation is the lack of validation in geographically independent cohorts with complete mortality follow-up; temporal validation within NHANES and preliminary external validation in a hospital-based Chinese cohort (n = 261) supported generalizability for identifying high-risk CKM status. Associations are not causes.
Based on the study abstract; the full paper, tables and supplementary data were not reviewed.
Source & supporting passages
Zhu Z, Ren K, Wang D, Lv Y, Jin H, Zhang L, Wang Y. · europepmc.org · October 6, 2026