What question did this study set out to answer?

This study aims to investigate the association between the triglyceride-glucose-body shape index (TyG-ABSI) and stroke risk.

March 5, 2026European journal of medical research2 citationsOpen Access

The value of triglyceride-glucose-body shape index (TyG-ABSI) in evaluating stroke: a multi-centre cross-sectional study

HLHanlin LiuShenyang Agricultural University MLMingxiao LiChina-Japan Friendship Hospital JHJinkui HaoNorthwestern University

Key Points

This study aims to investigate the association between the triglyceride-glucose-body shape index (TyG-ABSI) and stroke risk.
Conducted a multi-centre cross-sectional study with 25,263 participants from NHANES and CHARLS.
Utilized multivariable logistic regression to assess the association of TyG-ABSI with stroke.
Implemented subgroup analyses, restricted cubic spline regression, and threshold effect analyses for further evaluation.
Applied Pearson correlation analyses and Boruta’s algorithm for feature selection.
Evaluated TyG-ABSI's predictive ability using seven machine learning models and the SHAP algorithm.
A positive association was found between TyG-ABSI and stroke, with Odds Ratios of 1.186 in NHANES and 1.189 in CHARLS.
The association was stronger in non-elderly individuals without hypertension in NHANES.
The Gradient Boosting Machine model showed the best performance with an AUC of 0.784 in NHANES and 0.651 in CHARLS.
RCS regression indicated a linear relationship between TyG-ABSI and stroke, with no threshold effect.

Abstract

Stroke is the second leading cause of disability and mortality worldwide. A body shape index (ABSI) is calculated as waist circumference / (body mass index 2/3 × height 1/2). The TyG-ABSI is calculated as the product of the triglyceride-glucose index (TyG) and ABSI. To date, no studies have investigated the relationship between TyG-ABSI and stroke. Therefore, the purpose of this study is to fill this gap. This multi-centre cross-sectional study included 14,394 participants from the National Health and Nutrition Examination Survey (NHANES) cycles 2007–2018 and 10,869 participants from the China Health and Retirement Longitudinal Study (CHARLS) performed from 2015–2016. The potential association between TyG-ABSI and stroke was investigated using multivariable logistic regression, supplemented by subgroup analyses, restricted cubic spline (RCS) regression and threshold effect analyses. Pearson correlation analyses and Boruta’s algorithm were applied to select features. Subsequently, the discriminative ability of TyG-ABSI for stroke was assessed using seven machine learning models and the SHAP algorithm. After adjusting for sufficient covariates, the Odds Ratio (OR) for TyG-ABSI was 1.186 (95% CI: 1.027–1.370) in NHANES and 1.189 (95% CI: 1.058–1.336) in CHARLS. Subgroup analyses revealed that this association was more pronounced in non-elderly individuals without hypertension in NHANES. The RCS regression showed a linear relationship between TyG-ABSI and stroke, but no threshold effect was observed in two populations. Among the seven machine learning models, the Gradient Boosting Machine (GBM) performed the best, with an Area Under the Curve (AUC) of 0.784 (95%CI:0.751 − 0.817) in NHANES and 0.651(95%CI:0.604 − 0.699) in CHARLS. This study demonstrates a positive association between TyG-ABSI and stroke. The results suggest that TyG-ABSI is a relatively valuable indicator associated with stroke and can improve the development of public health policies.

问 AI

Bookmark

View Full Paper