Association between TyG-VAI and hypertension in children and adolescents based on machine learning algorithms

Hypertension in children and adolescents is increasingly prevalent, and effective risk indicators are needed. The triglyceride-glucose index combined with visceral adiposity index (TyG-VAI) has proved useful in adults, but evidence in the paediatric population remains limited. In this cross-sectional survey, 27,991 children and adolescents were enrolled in Henan Province, China, between 2023 and 2024 using multi-stage sampling. Multivariate logistic regression, subgroup and mediation analyses, and 11 machine learning models were used to examine the association between TyG-VAI and hypertension and to assess its predictive value. TyG-VAI was positively associated with hypertension (odds ratio [OR] = 2.36 for the highest quartile, P < 0.001), with greater sensitivity in males. Waist-to-height ratio was the main mediator (62.2%), followed by uric acid and the creatinine index. The multi-layer perceptron model incorporating TyG-VAI showed the best predictive performance (area under the curve [AUC] = 0.7518). TyG-VAI was independently associated with hypertension in children and adolescents and improved risk prediction, providing a basis for early intervention in this population.