Radiomics of Pericoronary Adipose Tissue and CT-FFR to Predict Major Adverse Cardiovascular Events in Patients with...

This study aims to integrate lesion-specific pericoronary adipose tissue (PCAT) radiomics analysis with existing clinical and imaging methods under the guidance of CT-derived fractional flow reserve (CT-FFR), to develop and validate an interpretable machine learning (ML) prediction model for patients with type 2 diabetes complicated by coronary artery disease (CAD). The performance of ML algorithms across different predictive models was compared using the area under the receiver operating characteristic curve (AUC). In the validation cohort, the XGBoost algorithm within the combined model achieved an AUC value of 0.908, outperforming the best algorithm in the traditional model (AUC = 0.834) and radiomics model (AUC = 0.840). Meanwhile, the Shapley algorithm highlights the additional incremental value of radiomic features. Our model enhances the predictive ability and provides clinicians with a comprehensive tool, facilitating early intervention for high-risk individuals and proactive secondary prevention strategies, which may potentially improve clinical outcomes.