Predicting outcomes following open revascularization for aortoiliac occlusive disease using machine learning
Ben Li1, Raj Verma2, Derek Beaton3
1Department of Surgery, University of Toronto, Toronto, ON, Canada; Division of Vascular Surgery, St. Michael's Hospital, Unity Health Toronto, University of Toronto, Toronto, ON, Canada; Institute of Medical Science, University of Toronto, Toronto, ON, Canada; Temerty Centre for Artificial Intelligence Research and Education in Medicine (T-CAIREM), University of Toronto, Toronto, ON, Canada.
Journal of Vascular Surgery
|July 16, 2023
Summary
Machine learning models accurately predict 30-day outcomes for open aortoiliac revascularization, outperforming logistic regression. These tools can guide risk mitigation strategies to improve patient results.
Area of Science:
- Vascular Surgery
- Machine Learning in Medicine
- Outcome Prediction
Background:
- Open surgical treatment for aortoiliac occlusive disease presents significant perioperative risks.
- Existing outcome prediction tools for these procedures are limited.
Purpose of the Study:
- To develop and validate machine learning (ML) algorithms for predicting 30-day outcomes after open aortoiliac revascularization.
- To compare the performance of ML models against traditional logistic regression.
Main Methods:
- Utilized the National Surgical Quality Improvement Program (NSQIP) vascular database (2011-2021).
- Trained six ML models using 38 preoperative variables, with data split into training (70%) and testing (30%) sets.
- Evaluated models using area under the receiver operating characteristic curve (AUROC) and calibration plots.
Main Results:
- The best performing model, XGBoost, achieved an AUROC of 0.95 for predicting 30-day major adverse limb event (MALE) or death.
- XGBoost significantly outperformed logistic regression (AUROC 0.79).
- Model performance was robust across various patient subgroups, with chronic limb-threatening ischemia identified as the strongest predictor.
Conclusions:
- Developed ML models accurately predict 30-day outcomes following open aortoiliac revascularization using preoperative data.
- These models demonstrate superior performance compared to logistic regression.
- The algorithms hold potential utility in guiding risk-mitigation strategies to improve patient outcomes.
