Graft restenosis risk prediction after coronary artery bypass surgery based on both flow and geometric configuration

Graft restenosis remains a significant challenge in coronary artery bypass grafting (CABG). Traditional function assessments, primarily relying on blood flow rate, often fail to capture the geometric and hemodynamic influences on graft patency. To address these limitations, this retrospective study aimed to establish a comprehensive risk prediction model that incorporates both flow dynamics and geometric features, facilitating clinically applicable evaluations. A total of 110 patient-specific CABG geometries were reconstructed from coronary computed tomography angiography (CCTA) images to extract key geometric parameters for subsequent statistical analysis. An additional 28 cases were analyzed for statistical and hemodynamic validation. Three logistic regression models were built and validated for restenosis risk prediction. Computational fluid dynamics (CFD) simulations were performed to investigate the hemodynamic characteristics of high-risk grafts. A MATLAB-based software tool was also developed to automate the analysis workflow. Among the three prediction models, the one combining graft flow and geometric factors balanced sensitivity and specificity, and performed best in the validation cohort (area under curve = 0.758, sensitivity = 89.1 %). CFD simulations on the validation cohort confirmed that grafts with high predicted risk exhibited poor hemodynamic conditions, including low time-averaged wall shear stress, high oscillatory shear index, and high relative residence time. Further statistical analysis revealed complex context-dependent interactions between graft flow and geometry. This study presents an integrated approach to restenosis risk prediction by combining patient-specific flow and geometric features. These findings are expected to enhance clinical decision-making and support more individualized postoperative management strategies in CABG.