Cross-population validation of the TyG-ABSI index as a novel predictor for chronic obstructive pulmonary disease: an
1Department of Emergency Critical Care Medicine, The Second Hospital of Hebei Medical University, Shijiazhuang, China. zhangjunli323@sina.com.
Background:
Chronic Obstructive Pulmonary Disease (COPD) is increasingly recognized as a systemic inflammatory condition closely linked to metabolic dysfunction. While traditional obesity metrics often fail to capture visceral adiposity and insulin resistance simultaneously, the Triglyceride-glucose body shape index (TyG-ABSI) has emerged as a composite marker of cardiometabolic risk. This study aimed to evaluate the association between TyG-ABSI and COPD risk and to assess its predictive utility using machine learning algorithms.
Methods:
Data were analyzed from two large-scale distinct cohorts: the China Health and Retirement Longitudinal Study (CHARLS, N = 2,771) and the National Health and Nutrition Examination Survey (NHANES, N = 1,925 pre-diabetic participants). The relationship between TyG-ABSI and COPD was assessed using multivariable logistic regression, subgroup analyses, and Restricted Cubic Splines (RCS). Furthermore, five machine learning models (XGBoost, LightGBM, CatBoost, Random Forest, and KNN) were developed to predict COPD risk. Model interpretability was achieved using SHapley Additive exPlanations (SHAP).
Results:
In both cohorts, elevated TyG-ABSI levels were significantly associated with an increased risk of COPD. Fully adjusted regression models revealed a linear dose-response relationship, with the highest quartile of TyG-ABSI showing a 59% increased risk in CHARLS (OR: 1.591) and a four-fold risk increase in NHANES (OR: 4.017) compared to the lowest quartile. Tree-based machine learning models, particularly XGBoost and LightGBM, demonstrated superior discriminative performance with AUCs exceeding 0.96 in the CHARLS dataset. SHAP analysis consistently identified TyG-ABSI as a high-priority feature contributing to disease prediction, irrespective of demographic variations.
Conclusions:
The TyG-ABSI index serves as a robust, independent risk factor for COPD across Chinese and American populations. Its integration into screening protocols, augmented by explainable machine learning, offers a valuable tool for early identification of high-risk individuals, highlighting the critical intersection of metabolic health and respiratory function.
