Suppr超能文献

整形外科学教育与培训中人工智能的现状:一项系统综述。

The Current Landscape of Artificial Intelligence in Plastic Surgery Education and Training: A Systematic Review.

作者信息

Genovese Ariana, Borna Sahar, Gomez-Cabello Cesar A, Haider Syed Ali, Prabha Srinivasagam, Trabilsy Maissa, Forte Antonio Jorge

机构信息

Division of Plastic Surgery, Mayo Clinic, Jacksonville, Florida.

Division of Plastic Surgery, Mayo Clinic, Jacksonville, Florida; Center for Digital Health, Mayo Clinic, Rochester, Minnesota.

出版信息

J Surg Educ. 2025 Aug;82(8):103519. doi: 10.1016/j.jsurg.2025.103519. Epub 2025 May 15.

Abstract

OBJECTIVE

Artificial intelligence (AI) shows promise in surgery, but its role in plastic surgery education remains underexplored. This review evaluates the current landscape of AI in plastic surgery education.

DESIGN

A systematic search was conducted on August 11, 2024, across PubMed, CINAHL, IEEE, Scopus, Web of Science, and Google Scholar using terms related to AI, plastic surgery, and education. Original research articles focusing on AI in plastic surgery education were included, excluding correspondence, reviews, book chapters, theses, corrections, and non-peer-reviewed or non-English articles. Two investigators independently screened studies and synthesized data. ROBINS-I was used to assess bias.

RESULTS

Fifteen studies were included, with 13 evaluating large language models (LLMs) such as ChatGPT, Microsoft Bing, and Google Bard. ChatGPT-4 outperformed other models on In-Service Examinations (average score of 72.7%) and demonstrated potential as a teaching assistant in plastic surgery education. AI-generated personal statements were comparable to human-written ones. However, ChatGPT showed inaccuracies in generating surgical protocols. ChatGPT demonstrated its ability to provide qualitative predictions, forecasting survey results that indicated limited current use of AI in plastic surgery education but support for further AI research. a study combined ChatGPT with DALL-E 2, a generative model, to create acceptable educational images. Machine learning was used in 1 study for evaluating surgical skill and providing real-time feedback during liposuction. Nine studies had low risk of bias, while 6 had moderate risk.

CONCLUSIONS

AI demonstrates potential as an educational tool in plastic surgery. However, limitations of evidence, such as AI model uncertainties, introduce ambiguity. While AI cannot replicate the expertise of seasoned surgeons, it shows promise for foundational learning and skill assessment. Developing authenticity guidelines and enhancing AI capabilities are essential for its effective, ethical integration into plastic surgery education.

摘要

目的

人工智能(AI)在外科手术中显示出应用前景,但其在整形外科学教育中的作用仍未得到充分探索。本综述评估了人工智能在整形外科学教育中的当前状况。

设计

于2024年8月11日在PubMed、CINAHL、IEEE、Scopus、科学网和谷歌学术上进行了系统检索,使用了与人工智能、整形外科学和教育相关的术语。纳入了专注于人工智能在整形外科学教育中的原创研究文章,排除了通信、综述、书籍章节、论文、勘误以及非同行评审或非英文文章。两名研究人员独立筛选研究并综合数据。使用ROBINS - I评估偏倚。

结果

纳入了15项研究,其中13项评估了诸如ChatGPT、微软必应和谷歌巴德等大语言模型(LLM)。ChatGPT - 4在内科医师在职考试中的表现优于其他模型(平均得分72.7%),并在整形外科学教育中展现出作为教学助手的潜力。人工智能生成的个人陈述与人工撰写的相当。然而,ChatGPT在生成手术方案时存在不准确之处。ChatGPT展示了其提供定性预测的能力,预测调查结果表明目前人工智能在整形外科学教育中的应用有限,但支持进一步的人工智能研究。一项研究将ChatGPT与生成模型DALL - E 2相结合,以创建可接受的教育图像。1项研究使用机器学习评估手术技能并在抽脂过程中提供实时反馈。9项研究的偏倚风险较低,而6项研究的偏倚风险为中等。

结论

人工智能在整形外科学中显示出作为教育工具的潜力。然而,证据的局限性,如人工智能模型的不确定性,带来了模糊性。虽然人工智能无法复制经验丰富的外科医生的专业知识,但它在基础学习和技能评估方面显示出前景。制定真实性指南并增强人工智能能力对于将其有效、合乎道德地融入整形外科学教育至关重要。

文献AI研究员

20分钟写一篇综述,助力文献阅读效率提升50倍。

立即体验

用中文搜PubMed

大模型驱动的PubMed中文搜索引擎

马上搜索

文档翻译

学术文献翻译模型,支持多种主流文档格式。

立即体验