ABSTRACT
INTRODUCTION: Malignant cerebral edema (MCE) is a serious complication and the main cause of poor prognosis in patients with large-hemisphere infarction (LHI). Therefore, the rapid and accurate identification of potential patients with MCE is essential for timely therapy. This study utilized an artificial intelligence-based machine learning approach to establish an interpretable model for predicting MCE in patients with LHI. METHODS: This study included 314 patients with LHI not undergoing recanalization therapy. The patients were divided into MCE and non-MCE groups, the extreme Gradient boosting (XGBoost) model was developed. A confusion matrix was used to measure the prediction performance of the XGBoost model. We also utilized the SHapley Additive extension (SHAP) method to explain the XGBoost model. Decision curve analysis and receiver operating characteristic (ROC) curve were performed to evaluate the net benefits of the model. RESULTS: MCE was observed in 121(38.5%) of the 314 patients with LHI. The model showed excellent predictive performance, with an area under the curve of 0.916. The SHAP method revealed the top 10 predictive variables of the MCE such as ASPECTS score, NIHSS score, CS score, APACHE II score, HbA1c, AF, NLR, PLT, GCS and Age based on their importance ranking. CONCLUSION: An interpretable predictive model can increase transparency and help doctors accurately predict the occurrence of MCE in LHI patients, not undergoing recanalization therapy within 48h from onset, providing patients with better treatment strategies and enabling optimal resource allocation.
ABSTRACT
Background: Malignant cerebral edema (MCE) is a life-threatening complication of large hemisphere infarction (LHI). Therefore, a fast, accurate, and convenient tool for predicting MCE can guide triage services and facilitate shared decision-making. In this study, we aimed to develop and validate a nomogram for the early prediction of MCE risk in acute LHI involving the anterior circulation and to understand the potential mechanism of MCE. Methods: This retrospective study included 312 consecutive patients with LHI from 1 January 2019 to 28 February 2023. The patients were divided into MCE and non-MCE groups. MCE was defined as an obvious mass effect with ≥5 mm midline shift or basal cistern effacement. Least absolute shrinkage and selection operator (LASSO) and logistic regression were performed to explore the MCE-associated factors, including medical records, laboratory data, computed tomography (CT) scans, and independent clinic risk factors. The independent factors were further incorporated to construct a nomogram for MCE prediction. Results: Among the 312 patients with LHI, 120 developed MCE. The following eight factors were independently associated with MCE: Glasgow Coma Scale score (p = 0.007), baseline National Institutes of Health Stroke Scale score (p = 0.006), Alberta Stroke Program Early CT Score (p < 0.001), admission monocyte count (p = 0.004), white blood cell count (p = 0.002), HbA1c level (p < 0.001), history of hypertension (p = 0.027), and history of atrial fibrillation (p = 0.114). These characteristics were further used to establish a nomogram for predicting prognosis. The nomogram achieved an AUC-ROC of 0.89 (95% CI, 0.82-0.96). Conclusion: Our nomogram based on LASSO-logistic regression is accurate and useful for the early prediction of MCE after LHI. This model can serve as a precise and practical tool for clinical decision-making in patients with LHI who may require aggressive therapeutic approaches.
