Predicting Short-Term Mortality after Endovascular Aortic Repair Using Machine Learning-Based Decision Tree Analysis.

Area of Science:

• Vascular Surgery
• Medical Informatics
• Artificial Intelligence in Medicine

Background:

• Endovascular aneurysm repair (EVAR) offers a less invasive option for abdominal aortic aneurysms.
• Identifying predictors of patient outcomes, especially for high-risk individuals, is critical for EVAR success.
• Machine learning (ML) techniques can uncover complex patterns in patient data to improve outcome prediction.

Purpose of the Study:

• To evaluate the efficacy of ML-based decision tree analysis (DTA) in predicting short-term mortality following EVAR.
• To identify key factors influencing patient outcomes in the context of EVAR.
• To leverage artificial intelligence for enhanced preoperative risk stratification in EVAR patients.

Main Methods:

• A DTA model was developed using Python 3.7 and the scikit-learn toolkit.
• The study analyzed data from 169 patients who underwent EVAR, examining 23 variables.
• The model aimed to predict short-term mortality (within 3 years) by identifying significant predictors.

Main Results:

• Poor nutritional status emerged as the primary predictor of short-term mortality after EVAR.
• Other significant predictors included chronic kidney disease, chronic obstructive pulmonary disease, and advanced age (octogenarian).
• The DTA model achieved 68.7% accuracy, 65.7% specificity, and 79.0% sensitivity, identifying 6 terminal nodes with mortality risks from 0% to 79.9%.

Conclusions:

• ML-based DTA shows significant promise for predicting short-term mortality after EVAR.
• The findings underscore the importance of thorough preoperative assessments, including nutritional status and comorbidities.
• Individualized management strategies informed by predictive analytics can optimize patient care and outcomes following EVAR.