Abstract
PURPOSE: Surgical site infection (SSI) is a major complication in patients with open hand injuries. However, current clinical risk assessment largely relies on subjective judgment or traditional scoring systems, which often lack predictive precision and generalizability. This study aimed to develop, compare, and externally validate multiple machine learning (ML) models for predicting SSI in open hand injuries using routinely collected clinical indicators. METHODS: A total of 800 patients with open hand injuries were retrospectively enrolled. The primary cohort (n=500) was randomly divided into training (70%, n=350) and internal testing (30%, n=150) sets, while an independent cohort (n=300) was used for external validation. Eight ML algorithms were constructed and compared, including logistic regression, decision tree, random forest, support vector machine, k-nearest neighbor, naive Bayes, extreme gradient boosting, and light gradient boosting machine. Model performance was evaluated using area under the curve (AUC), accuracy, sensitivity, specificity, and other metrics in internal cross-validation and external validation. SHapley Additive exPlanations (SHAP) were applied for feature interpretability. RESULTS: The random forest model demonstrated the best performance, with an AUC of 0.903 (95% CI 0.863 to 0.943) in training, 0.870 (95% CI 0.822 to 0.918) in internal testing, and 0.849 (95% CI 0.802 to 0.896) in external validation. Six key variables (age, smoking, diabetes mellitus, time from injury to surgery, wound contamination, and negative pressure drainage) were identified as the most influential predictors. SHAP analysis provided interpretable insights into their contributions to infection risk. CONCLUSION: The random forest model showed robust predictive performance and generalizability for SSI in open hand injuries. These findings highlight the model's potential as a clinical decision-support tool to assist surgeons in early risk stratification and personalized interventions, potentially reducing morbidity and improving outcomes. Future prospective studies are needed for further validation.