To explore the ability of machine learning methods to predict revascularization eligibility in patients with CAD(Coronary Artery Disease) and compare the efficacy of the XGBoost (Extreme Gradient Boosting) model combined with the SHAP ( Shapley Additive exPlanations) interpretability method against traditional models in revascularization screening. A retrospective analysis was conducted on 466 patients with confirmed or suspected CAD who were admitted to the First Hospital of Jilin University from January 2020 to May 2025, and the patients' imaging indicators were collected. The XGBoost model was constructed by integrating multi-dimensional indicators,optimized using 5-fold cross-validation, and combined with the SHAP method to quantify feature contribution. The results showed that the AUC(Area Under the Curve) of the XGBoost model reached 0. 899 (95% CI: 0. 871-0. 927), which was significantly higher than that of the traditional logistic regression model (AUC = 0. 812), the logistic model with full CCTA parameters (AUC = 0. 786). SHAP analysis identified minimum luminal area and maximum degree as the most critical predictors. The combination of XGBoost and SHAP can effectively assist in screening revascularization eligibility for CAD patients, with better predictive performance and interpretability than traditional models, providing reliable support for precise clinical intervention.

coronary artery disease ( CAD), revascularization eligibility, machine learning ( ML), coronary computed tomographic angiography(CCTA)