ChatGPT与Google在烧伤患者社会经济差异信息中的效用比较
本研究基于临床指南制定烧伤即刻处理、医疗治疗和长期护理的标准化问题,比较ChatGPT(v4.0)与Google首个搜索结果的准确性、可读性和社会经济地位(SES)相关性。2名医学生和2名烧伤外科医生用全局质量评分(GQS,1分差至5分优)评估准确性,并用Wilcoxon符号秩检验比较。ChatGPT的GQS为4.35±0.60,高于Google的2.25±1.10(P<.01);ChatGPT在74%的回答中涉及SES问题,Google为33%。
要点
- ChatGPT回答烧伤护理信息的质量评分高于Google(GQS 4.35±0.60 vs 2.25±1.10,P<.01)。
- 两平台阅读年级均为8和9级,但ChatGPT在74%回答中涉及费用与可及性等SES主题,Google为33%。
- ChatGPT在半数问题上被一致偏好,或有助于降低低SES患者的健康信息差异。
核心结论
- ChatGPT在提供准确且与SES相关的烧伤护理信息方面优于Google,可能有助于减少低SES患者的健康信息差异。
研究设计与样本
- 基于临床指南制定烧伤即刻处理、医疗治疗和长期护理的标准化问题,比较ChatGPT(v4.0)与Google首个搜索结果。
- 由2名医学生和2名烧伤外科医生使用GQS(1分差至5分优)评估准确性,以Flesch-Kincaid年级水平测量可读性,并统计涉及费用与可及性主题的回答评估SES相关性。
- 采用Wilcoxon符号秩检验比较准确性、可读性和SES相关性。
主要发现
- ChatGPT的GQS为4.35±0.60,高于Google的2.25±1.10(P<.01)。
- ChatGPT在半数问题上被一致偏好。
- 两平台阅读年级均为8和9级。
- ChatGPT在74%的回答中涉及SES问题,Google为33%。
关键数值与推荐
- 准确性采用GQS评分,范围1分(差)至5分(优)。
- 可读性采用Flesch-Kincaid年级水平。
- SES相关性通过计数包含费用与可及性相关主题的回答确定。
临床怎么用
- 临床可考虑向低SES烧伤患者推荐ChatGPT等AI工具获取准确且贴合费用与可及性需求的护理信息。
- 使用AI工具时仍需结合专业医疗判断,不应替代烧伤专科诊疗。
局限性
- 未来研究需在其他临床主题和患者人群中验证这些发现。
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
Patients from low-socioeconomic status (SES) backgrounds face barriers to quality burn care, such as limited healthcare access and follow-up. Many turn to online resources like Google, which may provide overwhelming or irrelevant information. This study compares the accuracy, readability, and SES-relevance of burn care information from ChatGPT and Google to address these disparities. A standardized set of questions on immediate burn care, medical treatments, and long-term care was developed based on clinical guidelines. Responses from ChatGPT (v4.0) and the first Google search result were analyzed. Two medical students and 2 burn surgeons assessed accuracy using the Global Quality Score (GQS) on a scale of 1 (poor) to 5 (excellent). Readability was measured using the Flesch-Kincaid grade level, and SES relevance was determined by counting responses that included themes related to affordability and access to care. Accuracy, readability, and SES relevance were then compared using a Wilcoxon signed-rank test. ChatGPT provided higher-quality responses (GQS 4.35 ± 0.60) than Google (GQS 2.25 ± 1.10, P < .01). ChatGPT was unanimously preferred for half of the questions. Both platforms had reading grade levels of 8 and 9, but ChatGPT addressed SES issues in 74% of responses, compared to Google's 33%. ChatGPT outperformed Google in providing accurate, SES-relevant burn care information. Artificial intelligence tools like ChatGPT may help reduce health information disparities for low-SES patients by offering tailored and user-friendly guidance. Future studies should validate these findings across other clinical topics and patient populations.
引用
- Beohon B, Lewis JE, Nguyen P, et al. Evaluating ChatGPT's Utility in Addressing Socioeconomic Disparities in Burn Patients: A Comparative Study With Google. J Burn Care Res. 2026. PMID: 40801276. doi:10.1093/jbcr/iraf158
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。