Aguilar Javier Suárez, Florido Manuel Viñuela, Mayol Julio, Cristóbal Lara, Maldonado Andrés A
From the Department of Plastic and Reconstructive Surgery, Hospital Universitario de Getafe, Madrid, Spain.
Hospital Clínico San Carlos, Instituto de Investigación Sanitaria San Carlos, Universidad Complutense de Madrid, Madrid, Spain.
Ann Plast Surg. 2025 May 1;94(5):512-515. doi: 10.1097/SAP.0000000000004232. Epub 2025 Jan 24.
Carpal tunnel syndrome (CTS) is the most common peripheral nerve entrapment disease, and it is a subject of great interest and concern to medical professionals and the general public. Our study aims to analyze and compare the quality and accuracy of the information related to CTS provided by social media platforms (SMPs) and the new large language models (LLM).
On YouTube, the first 20 videos in English and the first 20 videos in Spanish when searching for "carpal tunnel syndrome" and "síndrome túnel carpo" were selected. On Instagram, the first 20 videos with the hashtag #carpaltunnelsyndrome and #tunelcarpiano were chosen (in total 80 videos). Duration, number of likes, number of views, number of followers, upload date, and author category (medical specialist, patient, etc) were evaluated. Three specific questions about CTS were asked to 2 new LLMs (ChatGPT and Google Bard). The quality of information was analyzed and compared by two independent board-certified plastic surgeons using the Journal of American Medical Association (JAMA) and DISCERN scales.
LLMs showed a significant higher quality of information when compared with SMPs based on the DISCERN scores ( P < 0.05). Average DISCERN scores for answers given by ChatGPT and Google Bard were 52.83 and 57.83, respectively (good quality). In YouTube and Instagram, the average score for the 80 videos based on the JAMA scale was 1.92 (low reliability) and 25.18 (very low quality) on the DISCERN scale. Videos created by medical professionals in SMPs were associated with a higher JAMA and DISCERN scores ( P < 0.05). 53.8% of the videos were made by a nonmedical author.
The quality of information from LLMs was good and significantly better than in SMP. A low participation of board-certified surgeons in SMP was found. Board-certified surgeons should be more involved in LLM and SMPs to increase leadership, improve education, and spread knowledge of peripheral nerve surgery.
腕管综合征（CTS）是最常见的周围神经卡压性疾病，是医学专业人员和公众非常感兴趣和关注的课题。我们的研究旨在分析和比较社交媒体平台（SMPs）和新型大语言模型（LLM）提供的与腕管综合征相关信息的质量和准确性。
在YouTube上，搜索“腕管综合征”和“síndrome túnel carpo”时，分别选取前20个英文视频和前20个西班牙文视频。在Instagram上，选择带有#腕管综合征和#tunelcarpiano标签的前20个视频（共80个视频）。对视频时长、点赞数、观看数、关注者数、上传日期和作者类别（医学专家、患者等）进行评估。向两个新型大语言模型（ChatGPT和谷歌巴德）提出三个关于腕管综合征的具体问题。由两位独立的获得委员会认证的整形外科医生使用《美国医学会杂志》（JAMA）和DISCERN量表对信息质量进行分析和比较。
根据DISCERN评分，与SMPs相比，大语言模型显示出显著更高的信息质量（P<0.05）。ChatGPT和谷歌巴德给出答案的平均DISCERN评分为分别为52.83和57.83（质量良好）。在YouTube和Instagram上，基于JAMA量表，80个视频的平均评分为1.92（可靠性低），基于DISCERN量表的评分为25.18（质量非常低）。SMPs中由医学专业人员创建的视频与更高的JAMA和DISCERN评分相关（P<0.05）。53.8%的视频由非医学作者制作。
大语言模型的信息质量良好，且明显优于社交媒体平台。发现获得委员会认证的外科医生在社交媒体平台中的参与度较低。获得委员会认证的外科医生应更多地参与到大语言模型和社交媒体平台中，以增强领导力、改善教育并传播周围神经外科知识。
