Abstract 4147211: Machine Learning Models for Predicting In-Hospital Mortality in Acute Aortic Dissection Patients with Malperfusion

Introduction: Acute aortic dissection (AAD) remains a critical cardiovascular emergency with high in-hospital mortality, especially in patients with malperfusion syndromes. Accurate prediction of in-hospital mortality in these patients is crucial for improving clinical decision-making and patient outcomes. This study evaluates the performance of various machine learning (ML) models in predicting in-hospital mortality among AAD patients with malperfusion. Methods: The study included 451 acute aortic dissection with malperfusion from Henan Provincial Chest Hospital from August 2020 to September 2023. The dataset included demographic, clinical, laboratory results, and treatment strategies. Multiple ML models were developed and validated, including k-Nearest Neighbor (KNN) and Extreme Gradient Boosting Machine (XGB). Model performance was assessed using sensitivity, specificity, and the area under the receiver operating characteristic curve (AUC-ROC). Results: Our results demonstrated that the XGB model achieved the higher AUC-ROC (0.75, 95% CI: 0.73, 0.78), indicating excellent discriminative ability. KNN model showed lower AUC-ROC (0.66, 95% CI: 0.63, 0.68) but robust performance. Conclusions: XGB models offered promising tools for predicting in-hospital mortality in AAD patients with malperfusion. These models can assist clinicians in identifying high-risk patients and tailoring treatment strategies to improve survival outcomes.