An explainable machine learning model predicts 30-day readmission after vertebral augmentation

Osteoporotic vertebral compression fracture (OVCF) patients face high 30-day readmission risks after vertebral augmentation procedures (VAPs). Using electronic health records (EHRs) of 3,947 OVCF patients who underwent VAPs (2019-2024), we developed an interpretable machine learning model to identify readmission predictors. Eight algorithms were evaluated via 10-fold cross-validation, and XGBoost showed the best performance (area under the curve [AUC], sensitivity, specificity, F1 score, and decision curve analysis). SHapley Additive exPlanations (SHAPs) analysis revealed key predictors including frailty, fall history, prolonged hospitalization, comorbidities (pulmonary/kidney disease), advanced age, and hypoalbuminemia. A clinical web application was created for real-time risk stratification, visualizing individualized risk contributions via SHAP to enable proactive interventions and targeted prevention, thereby improving outcomes and reducing healthcare burden.