Article record

大型语言模型生成的常见鼻科疾病陈述和参考文献的准确性、完整性和幻觉。

Accuracy, completeness and hallucinations in large language model generated statements and references for common rhinologic disorders.

AI/ML鼻科IF 4.9Q1

文献信息

PMID	42873484
原文	在 PubMed 查看原文 ↗
发表日期	2026
作者	I Margulis 查看其他 7 位作者 M A M Slim；E C Sommer；Y AAlammar；A B Kassam；S Khalife；L Macias-Valle；D D Sommer
作者单位	Division of Otolaryngology-Head and Neck Surgery, Department of Surgery, McMaster University, Hamilton, ON, Canada. 查看其他 6 条单位信息 ENT Department, Forth Valley Royal Hospital, NHS Forth Valley, Scotland, United Kingdom.；School of Medicine, Dentistry and Nursing, Wolfson Medical School Building, Glasgow, United Kingdom.；Bachelor of Health Sciences Program, Wilfrid Laurier University, Waterloo, ON, Canada.；Department of Otolaryngology Head and Neck Surgery, King Saud University, Riyadh, Saudi Arabia.；Neurological Surgery, Intent Medical Group, Endeavor Health Advanced Neurosciences Institute, Northwest Community Hospital, Arlington Heights, IL, USA.；Department of Otolaryngology Head and Neck Surgery, La Salle University, Hospital Español de México, México City, México.
期刊	Rhinology
SCI 分区	Q1
IF	4.9
研究类型	AI/ML · 临床
所属专科	鼻科

中文摘要

大型语言模型（LLM）越来越多地被用于医学文档、临床决策和科学写作。尽管这些应用展示了令人印象深刻的能力，但对其准确性、虚构参考文献和伦理影响仍存在担忧（1-3）。大规模多任务语言理解（MMLU）测试被引入作为评估LLM在多样化任务中表现的综合性基准（4）。这项前瞻性比较两部分研究评估了ChatGPT（CGPT）-3.5生成的四种常见鼻科疾病（鼻出血、过敏性鼻炎、牙源性鼻窦炎和慢性鼻窦炎）的100词摘要的准确性和完整性，包括背景、检查和治疗，由国际鼻科学会成员评估。此外，我们评估了当代LLM版本（CGPT-5 thinking（5）、non-thinking（6）、Claude Sonnet 4.5 non-thinking（6）和Gemini 2.5（7））为相同提示生成的参考文献的准确性。耳鼻喉科医生使用标准化量表评估每项陈述的准确性、完整性和整体参考文献质量。

英文摘要

Large language models (LLMs) are increasingly adopted for medical documentation, clinical decision-making, and scientific writing. Although these applications demonstrated impressive capabilities, concerns persist regarding their accuracy, fabricated references, and ethical implications (1-3). The Massive Multitask Language Understanding (MMLU) test was introduced as a comprehensive benchmark to evaluate how well LLMs perform across a diverse range of tasks (4). This prospective comparative two part study evaluated the accuracy and completeness of ChatGPT (CGPT)-3.5-generated 100 word summaries of four common rhinologic disorders (epistaxis, allergic rhinitis, odontogenic sinusitis, and chronic rhinosinusitis) including background, investigations, and magement as assessed by international rhinology society members. Furthermore, we evaluated the accuracy of references generated by contemporary LLM versions for the same prompts (CGPT-5 thinking (5), non-thinking (6), Claude Sonnet 4.5 non-thinking (6) and Gemini 2.5 (7)). Otolaryngologists evaluated each statement using standardized scales for accuracy, completeness, and overall reference quality.