Liu, Y., et al. focused on developing a machine learning model that can help doctors accurately predict which stroke patients with a blocked basilar artery will do well after having a clot-removing procedure known as “endovascular treatment,” or EVT. The researchers looked at n=184 patients from Shanxi Provincial People’s Hospital, and collected 68 pieces of information, including stroke severity score, lab tests, and whether a patient needed a ventilator, among others. Three computer models – logistic regression, support vector machine (SVM), and LightGBM – were compared to see which model best predicted recovery following EVT. SHapley Additive explanations (SHAP) was used to assess which patient factors were most important in predicting recovery following EVT. The SVM model was shown to work the best, correctly predicting patient outcomes almost 90% of the time. The most important factors in recovery were the following: not requiring a ventilator, not needing a tracheotomy, lower stroke severity at admission, and lower blood creatinine. In summary, patients with milder strokes and fewer complications performed better after EVT – a finding confirmed through the SVM-SHAP machine learning model. The findings from this study demonstrate the use of machine learning models to predict surgical outcomes for stroke patients.