A machine learning model for predicting postoperative complication risk in young and middle-aged patients with femoral neck fractures

Objective: Femoral neck fractures are the most common type of hip fracture, and the postoperative complications associated with these fractures significantly affect patients' quality of life and healthcare costs. The objective of this study was to develop a predictive model using machine learning (ML) techniques to assess the risk of postoperative complications in young and middle-aged patients with femoral neck fractures. Methods: We retrospectively analyzed data from 899 young and middle-aged patients with femoral neck fractures who underwent surgical treatment between September 2019 and June 2024. Key predictors affecting postoperative complications were identified through LASSO regression and multifactorial logistic regression analyses. Several machine learning (ML) models were then integrated for comparative analysis. Ultimately, the best-performing model was selected, and its interpretation was provided using SHAP values to offer a personalized risk assessment. Results: The study results indicate that intraoperative reduction quality, medial cortex comminution, fracture types, posterior tilt angle, early postoperative weight-bearing, and removal of internal fixation devices are significant predictors of postoperative complications. The logistic regression model demonstrated the best performance on the test set, with an area under the curve (AUC) of 0.906, accuracy of 0.877, sensitivity of 0.748, and specificity of 0.903. Additionally, SHAP analysis identified the seven most important features in the model, providing clinicians with an intuitive tool for risk assessment. Conclusions: This study successfully developed and validated a logistic regression-based predictive model, augmented with SHAP explanations, providing an effective tool for assessing the risk of postoperative complications in young and middle-aged patients with femoral neck fractures.