Using Machine Learning to Predict Outcomes Following Thoracic and Complex Endovascular Aortic Aneurysm Repair

Area of Science:

• Vascular Surgery
• Medical Informatics
• Machine Learning in Healthcare

Background:

• Thoracic endovascular aortic repair (TEVAR) and complex endovascular aneurysm repair (EVAR) are high-risk procedures.
• Existing risk prediction tools for these interventions have limitations.
• There is a need for improved methods to predict patient outcomes.

Purpose of the Study:

• To develop and evaluate machine learning algorithms for predicting 1-year adverse outcomes after TEVAR and complex EVAR.
• To compare the performance of machine learning models against traditional logistic regression.

Main Methods:

• Utilized the Vascular Quality Initiative database (2012-2023) for patients undergoing elective TEVAR/EVAR.
• Extracted 172 preoperative, intraoperative, and postoperative features.
• Trained six machine learning models, including Extreme Gradient Boosting, using 70% of the data and validated on 30%.

Main Results:

• The Extreme Gradient Boosting model demonstrated superior predictive performance (AUC 0.96) compared to logistic regression (AUC 0.70) for preoperative prediction.
• The model maintained high accuracy at intraoperative (AUC 0.97) and postoperative (AUC 0.98) stages.
• Excellent calibration and low Brier scores indicated good agreement between predicted and observed outcomes.

Conclusions:

• Machine learning models, particularly Extreme Gradient Boosting, can accurately predict 1-year life-altering events after TEVAR and complex EVAR.
• These models offer improved performance over traditional logistic regression for outcome prediction.
• The findings support the use of machine learning in clinical decision-making for complex aortic repair procedures.