Prediction of Aneurysm Sac Shrinkage After Endovascular Aortic Repair Using Machine Learning-Based Decision Tree Analysis

Toshiya Nishibe; Tsuyoshi Iwasa; Seiji Matsuda; Masaki Kano; Shinobu Akiyama; Shoji Fukuda; Jun Koizumi; Masayasu Nishibe; Alan Dardik

doi:10.1016/j.jss.2024.11.049

Prediction of Aneurysm Sac Shrinkage After Endovascular Aortic Repair Using Machine Learning-Based Decision Tree Analysis

J Surg Res. 2025 Jan 9:306:197-202. doi: 10.1016/j.jss.2024.11.049. Online ahead of print.

Authors

Toshiya Nishibe¹, Tsuyoshi Iwasa², Seiji Matsuda², Masaki Kano³, Shinobu Akiyama³, Shoji Fukuda³, Jun Koizumi⁴, Masayasu Nishibe⁵, Alan Dardik⁶

Affiliations

¹ Faculty of Medical Informatics, Hokkaido Information University, Ebetsu, Japan; Department of Cardiovascular Surgery, Tokyo Medical University, Tokyo, Japan. Electronic address: toshiyanishibe@yahoo.co.jp.
² Faculty of Medical Informatics, Hokkaido Information University, Ebetsu, Japan.
³ Department of Cardiovascular Surgery, Tokyo Medical University, Tokyo, Japan.
⁴ Department of Radiology, Chiba University School of Medicine, Chiba, Japan.
⁵ Department of Surgery, Eniwa Midorino Clinic, Eniwa, Japan.
⁶ Department of Surgery, Yale School of Medicine, New Haven, Connecticut.

PMID: 39793306
DOI: 10.1016/j.jss.2024.11.049

Abstract

Introduction: A simple risk stratification model to predict aneurysm sac shrinkagein patients undergoing endovascular aortic repair (EVAR) for abdominal aortic aneurysms (AAA) was developed using machine learning-based decision tree analysis.

Methods: One hundred nineteen patients with AAA who underwent elective EVAR at Tokyo Medical University Hospital between November 2013 and July 2019 were included in the study. Predictors of aneurysm sac shrinkage identified in univariable analysis (P < 0.05) were entered into the decision tree analysis.

Results: Univariable analysis revealed significant differences between patients with and without aneurysm sac shrinkage in the variables of age (<75 y or ≥75 y), current smoking, operative type II endoleak, and preoperative pulse wave velocity (PWV) (<1800 cm/s or ≥1800 cm/s). The decision tree showed that preoperative PWV was the most relevant predictor, followed by operative type II endoleak and current smoking, and identified 6 terminal nodes with likelihoods of aneurysm sac shrinkage ranging from 5.6% to 63.6%.

Conclusions: We established a decision tree model with 3 variables (preoperative PWV, operative type II endoleak, and current smoking) to predict aneurysm sac shrinkage in patients undergoing EVAR for AAA. This classification model may help identify patients with a high or low likelihood of aneurysm sac shrinkage.

Keywords: Abdominal aortic aneurysm; Aneurysm sac shrinkage; Artificial intelligence; Decision tree analysis; Endovascular aortic repair; Machine learning.