Development and validation of a machine learning-based prediction model for emergence agitation in geriatric patients undergoing orthopedic surgery: a prospective cohort study

ObjectiveThis study aims to identify factors associated with emergence agitation in geriatric patients undergoing orthopedic surgery and to develop a machine learning model for predicting the risk of emergence agitation.MethodsThis prospective cohort study included 612 geriatric patients who underwent elective orthopedic surgery at the First Affiliated Hospital of Jinzhou Medical University in Liaoning Province between April 2025 and April 2026. The patients were randomly divided into a training set (n = 430) and a test set (n = 182) at a 7:3 ratio. Least absolute shrinkage and selection operator (LASSO) regression and multivariable logistic regression analysis were used to identify predictors. Six machine learning algorithms were employed to develop prediction models. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration curve, and decision curve analysis (DCA). The optimal model was interpreted using Shapley Additive Explanations (SHAP).ResultsThe incidence of emergence agitation among geriatric patients who underwent orthopedic surgery was 27.61%. Preoperative anxiety, preoperative cognitive function, dexmedetomidine use, intraoperative hypothermia, intraoperative blood loss, and postoperative NRS pain score were identified as key predictors. The random forest (RF) model demonstrated the best performance, achieving an AUC of 0.905, accuracy of 0.819, sensitivity of 0.820, specificity of 0.818, F1 score of 0.713, and Brier score of 0.111 on the test set. DCA showed that the model had the highest clinical net benefit. SHAP analysis revealed that intraoperative blood loss was the most significant predictor, followed by postoperative NRS pain score and preoperative cognitive function.ConclusionWe developed and internally validated six machine learning models to predict the risk of emergence agitation in geriatric patients undergoing orthopedic surgery. The RF model achieved the best performance. Following successful external validation in multicenter cohorts, the model may assist clinicians in identifying patients at increased risk of emergence agitation during early PACU recovery and in prioritizing intensified monitoring and timely supportive management.