Abstract
<title>Abstract</title> <p>Background The triglyceride–glucose (TyG) index and its derivatives (TyG–BMI, TyG–WC, and TyG–WHtR) have emerged as reliable markers. This study aims to evaluate their associations with hypertension and diabetes and determine their incremental predictive value using a machine learning approach. Method Data were obtained from 632 adult residents participating in the China Chronic Disease and Risk Factor Surveillance in Anji County. All statistical analyses were performed using R software (version 4.3.3). Continuous variables were expressed as mean ± standard deviation (SD) or median (interquartile range, IQR) and compared using the Wilcoxon rank-sum test or Kruskal-Wallis test. Categorical variables were presented as frequencies (percentages) and compared using the Chi-square test or Fisher's exact test.To assess the association between TyG-related indices and the risks of hypertension and diabetes, we employed both multivariate logistic regression and Cox proportional hazards regression models. Subsequently, a comprehensive machine learning workflow was implemented to predict disease risk and assess the incremental value of incorporating these indices.. Results The prevalence of both hypertension and diabetes showed a significant increasing trend across the quartiles of all TyG-related indices, which were all positively associated with both diseases. The Lasso model outperformed other machine learning algorithms in predictive accuracy. Incorporating these four indices into the basic model significantly improved discriminative ability for hypertension. For diabetes, integrating the TyG index, TyG-WC, and TyG-WHtR optimized risk stratification, whereas adding TyG-BMI yielded no significant improvement. Incorporating these indicators into machine learning models facilitates early risk identification and targeted clinical interventions. Conclusion The TyG index and its obesity-related derivatives are significant predictors of hypertension and diabetes. Incorporating these accessible biomarkers into machine learning frameworks provides substantial incremental predictive value, thereby optimizing early risk stratification and facilitating targeted clinical interventions.</p>