大语言模型生成放射报告患者摘要的前瞻性应用研究
本研究前瞻性纳入101例三级医院门诊影像患者,评估整合经放射科医生修订的大语言模型摘要、可点击词条及AI视频的网页应用对报告理解的影响。结果显示应用使用后中位理解评分显著提升,47.52%认为LLM摘要最有帮助,但多数患者对未经医生修订的LLM摘要持保留态度。
要点
- LLM摘要需放射科医生逐条修订,平均每篇删除24.75词,不能直接使用未经审核的机器输出
- 多数患者仅「有些放心」或「非常不放心」未经医生修订的LLM摘要,仍需专业把关
- 交互式材料中患者认为LLM摘要比可点击词条和AI视频更有帮助
核心结论
- 交互式LLM驱动的网页应用可显著提高患者对复杂放射报告的自报理解程度
- 患者对未经医生修订的LLM摘要存在明显顾虑,成功整合依赖专业审核,但增加的编辑工作量可能限制规模化落地
研究设计与样本
- 前瞻性研究,2025年5月至7月在一所三级大学医院门诊影像候诊区招募受试者
- 纳入101例,男40例,女61例,年龄20至82岁,平均58岁
- 检查完成且报告发布后,生成LLM患者友好摘要并经一名放射科医生审校准确性
- 受试者浏览包含经医生修订LLM摘要、可点击词条和AI视频的网页应用并完成调查
关键影像表现/征象
- LLM摘要经手工编辑后词汇相似度为84.63%,语义相似度为98.25%
- LLM摘要平均每篇删除24.75词,添加0.13词,说明临床可用性需逐句修订
- 47.52%的受试者认为LLM摘要最有帮助,优于可点击词条和AI视频
- 对未经医生编辑的LLM摘要,27.72%仅「有些放心」,25.74%「非常不放心」
诊断标准与关键数值
- 使用网页应用前理解评分中位数为4.00,使用后为5.00,P < .001
- 对未经医生编辑的LLM摘要感到放心程度的分布中,「有些放心」占27.72%,「非常不放心」占25.74%
临床应用与阅片要点
- 放射科医生在应用LLM摘要前必须逐篇审校,重点删除冗余且谨慎添加内容
- 向患者提供报告时可将LLM摘要与可点击词条、视频组合,优先突出文字摘要部分
- 报告解读材料需注明已经医生审核,否则患者信任度不足
局限性
- 单中心研究院内研究,样本量101例
- 理解评分为自报结果,非客观知识测验
- LLM摘要临床编辑增加放射科医生工作量,可能限制其在真实临床环境中的规模化推广
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)。
英文原文(摘要)· 点击展开
PURPOSE: To evaluate how a custom web application integrating clinician-edited large language model (LLM)-generated summaries, clickable definitions, and artificial intelligence (AI)-generated videos affects radiology report comprehension, feature preferences, and overall sentiment toward AI-assisted report summaries.
MATERIALS AND METHODS: This prospective study recruited participants between May and July 2025 at a hospital-based outpatient imaging floor before their scheduled examinations at a tertiary university hospital. After examination completion and report publication, patient-friendly AI summaries were generated and reviewed by a radiologist for accuracy. Participants were then shown a web application containing their own de-identified, AI-augmented reports featuring clinician-edited LLM-generated summaries with clickable terms and AI videos. Participants were surveyed on comprehension, feature usefulness, and attitudes toward LLM summaries.
RESULTS: Participants (n = 101, 40 male and 61 female, racially diverse subjects) ranged from 20 to 82 (mean 58 ± 15) years old. Overall comprehension improved significantly (median before using the web application: 4.00, median after: 5.00, P < .001), with 47.52% (n = 48) identifying LLM summaries as most helpful. However, LLM-summaries required manual clinician edits (average per summary: 24.75 words removed; 0.13 words added, lexical similarity = 84.63%; semantic similarity = 98.25%). When asked if they were comfortable with LLM summaries without clinician edits, most participants reported being only somewhat comfortable (27.72%) or very uncomfortable (25.74%).
CONCLUSION: This prospective study demonstrates that interactive, LLM-driven applications can significantly improve self-reported patient comprehension of complex radiology reports, emphasizing their potential to enhance patient-centered communication. However, patients had reservations about clinician-edited LLM-generated summaries, indicating that successful integration is contingent on professional oversight-an added workload that may limit scalable real-world implementation.
引用
- Li, Moon, Balthazar, et al. Enhancing Patient Understanding of Radiology Reports Through Large Language Model-Generated Summaries, Clickable Terms, and Artificial Intelligence Videos. J Am Coll Radiol. 2026. PMID: 42264206. doi:10.1016/j.jacr.2026.06.001
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。