Triangulating evidence for cardiometabolic Index: ROC cutoff, spline nonlinearity, and explainable machine learning...

Area of Science:

• Cardiovascular Health
• Metabolic Syndrome
• Public Health Screening

Background:

• The cardiometabolic index (CMI) integrates central adiposity and dyslipidemia.
• CMI may serve as a practical marker for community-based cardiovascular risk assessment.

Purpose of the Study:

• To evaluate the utility of the cardiometabolic index (CMI) in identifying high cardiovascular disease (CVD) risk.
• To assess CMI's performance in a Chinese cohort using conventional and machine learning methods.

Main Methods:

• Analysis of cross-sectional data from the ChinaHEART Luohe cohort (N=6,701).
• Cardiovascular disease (CVD) high risk defined by WHO criteria (10-year predicted risk ≥20%).
• Cardiometabolic index (CMI) evaluated using logistic regression, restricted cubic splines, and an explainable machine learning pipeline (LASSO-random forest-SHAP).

Main Results:

• Higher CMI (≥0.7) was associated with increased prevalence of cardiovascular disease (CVD) high risk (29% vs. 18%).
• Cardiometabolic index (CMI) alone showed modest discrimination (AUC: 0.571), but improved in multivariable models (AUC: 0.642).
• Machine learning models demonstrated strong predictive performance (Random Forest AUC: 0.814), with CMI ranked as a key feature.

Conclusions:

• Elevated cardiometabolic index (CMI) is significantly associated with cardiovascular disease (CVD) high-risk status.
• The association between CMI and CVD risk exhibits a nonlinear pattern.
• Findings support CMI's role as a valuable adjunct screening marker for cardiovascular risk.