Association between the triacylglycerol–glucose index, TyG-BMI and hypertension in children and adolescents: a cross-sectional study with machine-learning-based risk stratification

Background Hypertension in children and adolescents is closely linked to obesity and metabolic disturbance. The triacylglycerol–glucose (TyG) index and TyG-BMI are surrogate markers of insulin resistance, but their associations with pediatric hypertension remain incompletely characterized. Methods This cross-sectional study included 25,886 children and adolescents aged 7–17 years from 18 prefecture-level cities in Henan Province, China. The TyG index was calculated as ln[triglycerides (mg/dL) × fasting plasma glucose (mg/dL)/2] and TyG-BMI as TyG × BMI. Hypertension was defined as a blood pressure at or above the 95th percentile for age, sex and height (mean of the last two of three measurements). Logistic regression (Models I–IV) estimated odds ratios (ORs) per 1 SD and across quartiles, with non-linearity and effect modification assessed on both scales. Eleven machine-learning models were developed on a stratified 70:30 split, with imputation and feature selection performed strictly within the training set and discrimination (area under the receiver-operating characteristic curve, AUC) evaluated in the held-out validation set. Results Among 25,886 participants, 1,798 (6.95%) had hypertension. Median TyG was 8.05 (IQR 7.77–8.35) and median TyG-BMI 154 (IQR 133–180). Each 1-SD increment in TyG was associated with hypertension in all models (Model IV 1.306 (95% CI 1.145–1.489)), and the association was graded across quartiles; TyG-BMI showed stronger associations (Model IV 2.109 (95% CI 1.985–2.241)). No multiplicative product term was significant after correction, whereas on the additive scale all six relative excess risk estimates were positive with confidence intervals excluding zero. In the held-out validation set, discrimination was moderate (best AUC 0.7544, 95% CI 0.7302–0.7777, for TyG and 0.7534, 0.7298–0.7759, for TyG-BMI), and adding either index to the same baseline model already containing BMI increased discrimination only marginally (about 0.003 in AUC). Conclusions Higher TyG and TyG-BMI were associated with hypertension in Chinese children and adolescents, more strongly for TyG-BMI. No multiplicative interaction survived correction, but all additive-scale estimates were positive, so effect modification cannot be excluded. Machine-learning models built from routinely available variables showed only moderate discrimination and remain hypothesis-generating tools for risk stratification rather than validated prediction models. Longitudinal studies are needed to assess causality.