World Cardiology Conference 2026

Abstract

Importance: Bird-beak configuration of the stent-graft is responsible for reducing the effective proximal landing zone and increasing the risk of type Ia endoleak. and distal migration, collapse and fracture of the stent-graft. However, development of a machine-learning model based on aortic geometric features for Bird-beak configuration prediction after zone 2 thoracic endovascular aortic repair has not been reported, and which indicated the necessity of our research.

Objective: To develop a machine learning model for geometric modeling of thoracic aorta and risk prediction of bird-beak configuration during zone 2 thoracic endovascular aortic repair.

Design, setting and Participants: This retrospective cohort study was performed in a single tertiary medical center. From October 2019 to March 2022, 201 patients with acute type B aortic dissections who underwent zone 2 thoracic endovascular aortic repair were enrolled retrospectively. 131 patients were used as the training and internal validation cohorts, and 70 patients were selected as the external validation cohort.

Conclusions and Relevance Among the constructed multiparameter and 64-point-centreline-curvature models, the neural network had the best overall performance in predicting the risk of Bird-beak configuration occurrence, followed by random forest and decision tree. The model developed in this study is efficient and effective for risk stratification and planning thoracic endovascular aortic repair, and is easy to update for optimization.