Hallux Valgus Postoperative Satisfaction Prediction Algorithm Driven by Auto-Recursive Feature Elimination Ensemble Learning

In the field of hallux valgus, there has been increasing interest in establishing a postoperative satisfaction prediction model for hallux valgus that integrates multiple clinical variables. However, most existing studies on predicting patient satisfaction after hallux valgus surgery have focused on traditional statistical methods (such as logistic regression) or single machine learning models, exploring the impact of single or a few factors on satisfaction, and have failed to fully explore the interactions between variables. To address these issues, we propose a hallux valgus surgery satisfaction prediction model driven by Automatic Recursive Feature Elimination (Auto-RFE) and eXtreme Gradient Boosting (XGBoost) ensemble learning. Compared with previous research methods, when dealing with high-dimensional and imbalanced medical data analysis, this prediction model introduces the Synthetic Minority Over-sampling Technique (SMOTE) to solve the data imbalance problem, and Auto-RFE can dynamically optimize the optimal number of features, significantly improving the generalization ability of the prediction model. In addition, the model can effectively capture nonlinear relationships and interactions between features (such as the synergistic effect of pain and activity limitations) without expert experience, and the selected optimal feature set has higher prediction accuracy.