人工智能语言模型简化烧伤护理患者教育材料的可读性研究
本研究评估人工智能语言模型简化烧伤护理患者教育材料的可读性效果。结果显示ChatGPT 4o和Microsoft Copilot的可读性评分与传统计算器相当,而Microsoft Copilot、Gemini和Meta AI在简化文本时可将可读性评分降至理想的6至8年级水平。
要点
- ChatGPT 4o和Microsoft Copilot的可读性评分与传统可读性计算器相当
- Microsoft Copilot、Gemini和Meta AI可将文本可读性降至6至8年级水平
- AI简化材料仍需临床医生监督以确保准确性和适当性
核心结论
- AI模型(特别是Microsoft Copilot、Gemini和Meta AI)可有效简化医学文本,使其更易被患者理解,但需要临床医生监督以确保简化材料的准确性和适当性。
主要发现
- ChatGPT 4o和Microsoft Copilot的可读性评分与传统可读性计算器提供的评分相当。
- Microsoft Copilot、Gemini和Meta AI在简化文本任务中,将可读性评分降至6至8年级水平。
关键数值与推荐
- 全国指南推荐患者教育材料应编写在6至8年级阅读水平。
临床怎么用
- AI模型可用于简化烧伤护理患者教育材料,但临床医生需监督确保准确性和适当性,以促进健康素养和患者结局。
局限性
- 原文未述局限性
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
Health literacy is essential in patient care, especially in burn treatment, where understanding care information can significantly influence recovery outcomes. Despite national guidelines recommending that patient education materials be written at a 6th- to 8th-grade reading level, many resources exceed this complexity, exacerbating poor health literacy and patient outcomes. This study investigates the effectiveness of artificial intelligence language learning models in simplifying patient-facing burn care information to adhere to these readability standards. Fifteen excerpts from academic institutions' burn care materials were evaluated for readability using a traditional readability calculator and four AI models: ChatGPT 4o, Microsoft Copilot, Gemini, and Meta AI. The traditional readability calculator provided a baseline score, which was compared to the scores from the AI models. Results indicated that ChatGPT 4o and Microsoft Copilot had readability scores that were comparable to the readability scores provided by the traditional calculator. Additionally, when tasked with simplifying the texts, Microsoft Copilot, Gemini, and Meta AI reduced the readability scores to within the desired 6th to 8th-grade level. These findings suggest that AI models, particularly Microsoft Copilot, Gemini, and Meta AI, can effectively simplify medical texts, making them more accessible to patients. However, clinician oversight is necessary to ensure the accuracy and appropriateness of the simplified materials, promoting better health literacy and patient outcomes in burn care.
引用
- Vallurupalli M, Shah ND, Yadalla S, et al. Promoting patient health literacy in burn care through artificial intelligence language learning models: A study of text analysis and simplification. Burns. 2025. PMID: 40460491. doi:10.1016/j.burns.2025.107548
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。