Predicting Acute Stroke Risk Post-Carotid Artery Stenting Using Machine Learning and MRI Plaque Imaging

Motivation: Atherosclerosis is a major health issue in Japan, leading to complications like cerebral infarction after carotid artery stenting. Advances in machine learning offer potential for improved risk prediction. Goal(s): To evaluate whether machine learning applied to preoperative blood tests and MRI data can predict the risk of cerebral infarction complications post operation. Approach: Analyzed data from 175 patients, using 20 features, including clinical and imaging data. Employed Random Forest, Support Vector Machine, and Gradient Boosting algorithms with leave-one-out cross-validation, evaluating performance via AUC-ROC. Results: Random Forest achieved an AUC: 0.79, outperforming Gradient Boosting's AUC: 0.63, indicating effective risk prediction using machine learning. Impact: The study's results enable clinicians to better predict cerebral infarction risk post-carotid artery stenting using machine learning, potentially improving patient outcomes. It encourages further research into AI-driven risk assessment, enhancing personalized medicine and preoperative decision-making.