目的 构建并验证急诊颅脑损伤患者低血糖风险预测模型，为早期临床干预提供依据。方法 回顾性纳入2022年1月—2024年12月山东省某三级甲等医院急诊科收治的362例颅脑损伤患者为建模组，记录患者低血糖发生情况，通过单因素分析及多因素Logistic回归筛选低血糖危险因素，基于R语言构建列线图模型。采用受试者操作特征曲线（receiver operator characteristic curve，ROC）评估模型区分度，Hosmer-Lemeshow检验评价校准度。前瞻性收集2025年1月—5月同中心收治的90例颅脑损伤患者进行外部验证。结果 362例颅脑损伤患者中低血糖发生率为21.55%（78/362）。模型纳入格拉斯哥昏迷评分（Glasgow Coma Scale，GCS）、血糖变异系数、创伤严重度评分（Injury Severity Score，ISS）、合并糖尿病史、颅内压升高、使用胰岛素及营养支持等7项独立预测因子。建模组ROC曲线下面积（area under the curve，AUC）为0.852（95%CI：0.802~0.868），最佳截断值为0.403时，灵敏度为0.832，特异度为0.783。验证组AUC为0.847（95%CI：0.796~0.899），最佳截断值为0.418时，灵敏度为0.815，特异度为0.769。结论 该研究构建的列线图模型能有效识别急诊颅脑损伤低血糖高风险患者，有助于优化血糖监测策略及降低继发性脑损伤风险。

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.