Background: Respiratory Syncytial Virus (RSV) is a leading cause of severe lower respiratory tract infections in children, making early risk identification crucial for improving outcomes. This study aims to develop an interpretable machine learning model incorporating SHAP to predict severe RSV infection in children and identify key risk factors, providing an accurate and transparent tool for clinical decision-making. Methods: We retrospectively collected data from RSV-infected inpatients at Tianjin Children's Hospital (2019-2023). After predictor selection using LASSO regression, 14 features were included. Seven algorithms (logistic regression, random forest, XGBoost, support vector machine, GBDT, decision tree, and AdaBoost) were employed to develop prediction models. Model performance was evaluated using five-fold cross-validation and AUC. SHAP was applied to interpret the optimal model. Results: A total of 2595 pediatric inpatients with RSV infection were enrolled, including 2435 (93.8%) mild cases and 160 (6.2%) severe cases. Among the seven machine learning algorithms, the GBDT model achieved the best performance with an AUC of 0.8923 (95% CI: 0.845-0.931). SHAP analysis identified key predictors contributing to the model's decisions. Conclusion: The GBDT-based model effectively predicts severe RSV infection in children (AUC=0.8923), and SHAP enhances interpretability , offering a reliable tool for early identification of high-risk patients.