{"title":"Predicting Early Mortality after Thoracic Endovascular Aneurysm Repair: A Machine Learning-Based Decision Tree Analysis.","authors":"Masaki Kano, Toshiya Nishibe, Tsuyoshi Iwasa, Seiji Matsuda, Shinobu Akiyama, Toru Iwahashi, Shoji Fukuda, Yusuke Shimahara, Masayasu Nishibe","doi":"10.3400/avd.oa.25-00009","DOIUrl":null,"url":null,"abstract":"<p><p><b>Objectives:</b> Thoracic endovascular aneurysm repair (TEVAR) has revolutionized the treatment of thoracic aortic aneurysms (TAA) by providing a less invasive alternative to open surgery. This study aims to identify risk factors for early mortality following TEVAR for degenerative TAA using a machine learning-based decision tree analysis (DTA). <b>Methods:</b> This retrospective observational study analyzed 79 patients who underwent elective TEVAR to identify predictors of early mortality (within 2 years) using decision tree analysis. The dataset included 36 variables, covering age, sex, nutritional status, comorbidities, inflammation, immune status, and surgical details. The decision tree classifier was developed and validated using Python 3.7 with the scikit-learn toolkit. <b>Results:</b> DTA identified octogenarian status as the strongest predictor of early mortality, followed by poor nutritional status, debranching procedures, and compromised immunity. The model identified 7 terminal nodes, with early mortality risk ranging from 0% to 77.7%. It demonstrated moderate accuracy (65.8%) and high sensitivity (81.0%) but had relatively low specificity (60.3%), effectively identifying high-risk patients. <b>Conclusions:</b> Machine learning-based DTA identified key predictors of early mortality following TEVAR, including octogenarian status, poor nutritional status, compromised immunity, and debranching procedures. The model provides an interpretable risk stratification tool, but its clinical applicability requires further validation.</p>","PeriodicalId":7995,"journal":{"name":"Annals of vascular diseases","volume":"18 1","pages":""},"PeriodicalIF":0.6000,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12117201/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Annals of vascular diseases","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.3400/avd.oa.25-00009","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2025/5/23 0:00:00","PubModel":"Epub","JCR":"Q4","JCRName":"PERIPHERAL VASCULAR DISEASE","Score":null,"Total":0}
引用次数: 0
Abstract
Objectives: Thoracic endovascular aneurysm repair (TEVAR) has revolutionized the treatment of thoracic aortic aneurysms (TAA) by providing a less invasive alternative to open surgery. This study aims to identify risk factors for early mortality following TEVAR for degenerative TAA using a machine learning-based decision tree analysis (DTA). Methods: This retrospective observational study analyzed 79 patients who underwent elective TEVAR to identify predictors of early mortality (within 2 years) using decision tree analysis. The dataset included 36 variables, covering age, sex, nutritional status, comorbidities, inflammation, immune status, and surgical details. The decision tree classifier was developed and validated using Python 3.7 with the scikit-learn toolkit. Results: DTA identified octogenarian status as the strongest predictor of early mortality, followed by poor nutritional status, debranching procedures, and compromised immunity. The model identified 7 terminal nodes, with early mortality risk ranging from 0% to 77.7%. It demonstrated moderate accuracy (65.8%) and high sensitivity (81.0%) but had relatively low specificity (60.3%), effectively identifying high-risk patients. Conclusions: Machine learning-based DTA identified key predictors of early mortality following TEVAR, including octogenarian status, poor nutritional status, compromised immunity, and debranching procedures. The model provides an interpretable risk stratification tool, but its clinical applicability requires further validation.