目的 评估ChatGPT-4在外科疾病患者护理问题和护理措施自主回答的准确性，为护理教育的智能化改革提供参考。方法 选取某三级甲等医院2024年的30例外科住院患者病例为分析资料，按输入前训练、编写指令提示语、输入指令的步骤生成护理问题和护理措施，由5名护理专家根据统一的评分标准进行准确性评价，采用Fleiss’s Kappa检验和Cronbach’s α系数进行一致性检验。结果 30个案例护理问题和护理措施的准确性评为“非常好”分别占比85.3%、80.0%。每个病例的专家评分两两比较Fleiss’s Kappa检验值为0.433~0.763，所有病例护理问题和护理措施的总Cronbach’s α系数分别为0.908、0.943。结论 ChatGPT-4自主生成护理问题和护理措施的准确性整体可接受，反映出大语言模型具有成为教学辅助工具的潜力，但还需对其生成信息进一步判断，可为人工智能在护理教育领域的进一步应用和发展提供参考。

Objective To evaluate the accuracy of ChatGPT-4 in autonomously answering nursing questions and measures for surgical patients，and to provide a reference for the intelligent reform of nursing education.Methods Thirty inpatient cases from a tertiary first-class hospital in 2024 were selected as analytical data. Nursing problems and measures were generated following steps of pre-training input，writing instructional prompts，and inputting instructions. Five nursing experts evaluated the accuracy based on a unified scoring standard，and consistency was tested using Fleiss’s Kappa test and Cronbach’s α coefficient.Results The accuracy of nursing questions and measures for the 30 cases was rated as “excellent” at 85.3% and 80.0%，respectively. The Fleiss’s Kappa test values for pairwise comparison of expert scores for each case ranged from 0.433 to 0.763，and the overall Cronbach’s α coefficients for nursing questions and measures across all cases were 0.908 and 0.943，respectively.Conclusion The overall accuracy of autonomously generated nursing questions and measures by ChatGPT-4 is acceptable，reflecting the potential of large language models to serve as teaching aids. However，further judgment of the generated information is necessary，which can provide a reference for the further application and development of artificial intelligence in the field of nursing education.