Abstract
Diabetes mellitus with hypertension (DM + HTN) markedly elevates cardiovascular risks, yet its predictors remain unclear. Analyzing 5210 DM patients from three cohorts, this study identified serum cystatin C (CysC) as an independent risk factor for DM + HTN through univariate and multivariate logistic regression. A risk prediction model incorporating CysC concentration was developed and adjusted for age, sex, race, education, body mass index, smoking status, and drinking status. The model demonstrated good predictive performance and net benefit through receiver operating characteristic curves, calibration curves, and decision curve analysis. Restricted cubic spline analysis demonstrated a nonlinear relationship between CysC levels and DM + HTN risk, with concentrations above 0.94 mg/L exhibiting elevated risk. The model's performance was further evaluated using 10 machine learning algorithms and interpreted using SHapley Additive exPlanations (SHAP). This research provides a CysC-based model to aid clinicians in early identification of high-risk individuals.