Development of a Machine Learning-Based Prediction Model and an Online Risk Calculator for Preoperative Prediction of Pneumonia After Surgical Clipping for Aneurysmal Subarachnoid Hemorrhage

Background: Postoperative pneumonia (POP) is a frequent complication after microsurgical clipping for aneurysmal subarachnoid hemorrhage (aSAH). Early preoperative identification of high-risk patients may facilitate timely preventive management. This study aimed to develop and internally validate an explainable machine learning model based on routinely available admission variables and translate it into a web-based risk calculator. Methods: This single-center retrospective study included 574 patients with aSAH who underwent microsurgical clipping between 2019 and 2024. Patients were randomly divided into training and internal validation cohorts at a ratio of 7:3. Feature selection was performed using Spearman correlation analysis and the Boruta algorithm. Thirteen machine learning algorithms were compared using five-fold cross-validation, and the optimal model was evaluated in the internal validation cohort. SHAP analysis was used to interpret model predictions. Results: Thirteen predictors were retained for model development. The random forest model implemented using ranger showed the best overall performance, achieving an AUC of 0.887 (95% CI, 0.838–0.936) in the internal validation set. Calibration was acceptable, with a Brier score of 0.135, and decision curve analysis suggested potential clinical net benefit. SHAP analysis identified D-dimer and Glasgow Coma Scale score as the leading contributors to model predictions. An interactive R Shiny calculator was developed for individualized POP risk estimation. Conclusions: An explainable ranger-based random forest model using readily available admission and preoperative variables showed good discrimination and acceptable calibration for predicting POP after aSAH clipping. The model may enable risk estimation before surgery without specialized testing and support early individualized risk stratification. External and prospective validation is required before broader clinical implementation.