基于TyG指数联合预测冠心病的模型构建与验证

Model Construction and Validation for the Joint Prediction of Coronary Heart Disease Based on the TyG Index

摘要: 目的：构建基于甘油三酯葡萄糖指数(triglyceride-glucose index, TyG)联合预测冠状动脉粥样硬化性心脏病[以下简称为冠心病(coronary heart disease, CHD)]患病风险的预测模型，并验证该模型的预测性能。方法：选取2020年1月至2023年3月泰安市中心医院收治的已行冠状动脉造影检查的患者为建模研究对象，根据冠状动脉造影结果分为冠心病组(732例，67.97%)和非冠心病组(365例，33.03%)，比较两组一般资料及临床特征，采用多因素Logistic回归分析构建预测模型，绘制该预测模型的受试者工作特征(receiver operating characteristic, ROC)曲线及曲线下面积(area under the curve, AUC)分析评价该模型的预测效能，再通过选取2024年1月至2024年12月山东大学第二医院收治的已行冠状动脉造影检查的患者为验证模型的研究对象进行模型的外部验证。结果：TyG指数、年龄、纤维蛋白原是CHD发生的独立危险因素，既往是否服用阿司匹林是CHD发生的独立保护因素。男性更容易患CHD，该预测模型ROC为0.706，灵敏度为0.643，特异度为0.551。对该模型进行外部验证，模型验证的ROC曲线AUC为0.649 (95% CI: 0.611~0.687)，该模型具有良好的预测性能。结论：本研究构建的基于TyG指数联合性别、年龄、纤维蛋白原、服用阿司匹林史对CHD的发生具有良好的预测性能。

Abstract: Objective: To construct a prediction model based on the triglyceride-glucose index (TyG) for predicting the risk of coronary atherosclerotic heart disease (hereinafter referred to as coronary heart disease, CHD), and to verify the prediction performance of this model. Methods: Those who underwent coronary angiography at Tai’an Central Hospital from January 2020 to March 2023 were selected as the modeling study subjects. According to the results of coronary angiography, the patients were divided into the CHD group (732 cases, 67.97%) and the non-D group (365 cases, 33.03%). The general information and clinical characteristics of the two groups were compared. A prediction model was constructed using multivariate Logistic regression analysis, and the receiver operating characteristic (ROC) curve and the area under the curve (AUC) of the prediction model were drawn to evaluate prediction performance of the model. Then, the external validation of the model was performed by selecting patients who underwent coronary angiography at the Second Hospital of Shandong University from January 2024 to December 2024 as the study subjects for the validation model. Results: TyG index, age, and fibrinogen independent risk factors for the occurrence of CHD, and the history of taking aspirin was an independent protective factor for the occurrence of CHD. Men were more to suffer from CHD. The ROC of this prediction model was 0.706, the sensitivity was 0.643, and the specificity was 0.551. The ROC curve AUC of the model verification was 0.649 (95% CI: 0.61~0.687), and the model had good prediction performance. Conclusion: The prediction model based on TyG index combined with gender, age, fibrin, and history of taking aspirin for the occurrence of CHD had good prediction performance.