Abstract:

Objective To construct and validate a prediction model for hypoglycemia risk in emergency brain injury patients，providing a basis for early clinical intervention. Methods A total of 362 brain injury patients admitted to the emergency department of a tertiary hospital from January 2022 to December 2024 were retrospectively recruited as the modeling group，and hypoglycemia in patients was recorded. The screen for risk factors for hypoglycemia was analyzed by univariable analysis and multivariable logistic regression，and nomogram was constructed based on R language. The model’s discrimination，and the Hosmer-Lemeshow test were evaluated by receiver operating characteristic（ROC） curve. A prospective collection of 90 cases of patients admitted to the same center were conducted for external validation from January to May 2025. Results The rate of hypothermia among 362 brain injury patients was 21.55%（78/362）. The model included 7 predictive factors：Glasgow Coma Scale （GCS），coefficient of variation of blood glucose，Injury Severity Score（ISS），history of diabetes，elevated intracial pressure，use of insulin，and nutritional support. The ROC area under the curve（AUC） for the validation group was 0.852（95%CI：0.802-0.868），with sensitivity of 0.832 and specificity of 0.783 at the optimal cutoff value of 0.403. The AUC for the validation group was 0.847（95%CI：0.796-0.899），with a sensitivity of 0.815 and specificity of 0.769 at a cutoff value of 0.418. Conclusion The nomogram model constructed in this study can effectively identify high-risk groups for hypoglycemia in emergency brain injury patients，providing clinical guidance for optimizing blood glucose strategies and reducing the risk of secondary brain injury.

Key words: Emergency Brain Injuries, Hypoglycemia, Risk Prediction Model, Nursing