Search research articles

关于 JoVE

概览领导团队博客 JoVE 帮助中心

作者

出版流程编辑委员会范围与政策同行评审常见问题投稿

图书馆员

用户评价订阅访问资源图书馆顾问委员会常见问题

研究

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments 存档

教育

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual 教师资源中心教师网站

使用条款与条件

Search research articles

相关实验视频

Updated: Mar 1, 2026

Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application

Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application

Published on: April 14, 2023

皮肤学图像解释的大型语言模型 - - 一项比较研究.

Lasse Cirkel^1,2, Fabian Lechner^1,2, Lukas Alexander Henk^2,3

¹Institute of Artificial Intelligence, University Hospital Gießen-Marburg, Philipps University, Marburg, Germany.

Diagnosis (Berlin, Germany)

|May 27, 2025

概括

相关概念视频

您也可能阅读

相关文章

通过共同作者、期刊和引用图与本文相关的文章。

排序

Same author

Distinctive Reactivity to the C-terminal Epitope of BP180 Characterizes Immune Checkpoint Inhibitor-associated Bullous Pemphigoid, and an ELISA Based on the BP180 Ectodomain Enables Prompt Diagnosis in a Subset of Patients.

Acta dermato-venereologica·2026

Same author

Systemic sclerosis triggered by bleomycin: a 5-year follow-up.

EULAR rheumatology open·2026

Same author

VEXAS syndrome in a female with constitutional monosomy X.

EULAR rheumatology open·2026

Same author

Comparative analysis of large language models and traditional diagnostic decision support systems for rare rheumatic disease identification.

EULAR rheumatology open·2026

Same author

Bimekizumab rescue therapy in a patient with relapsing polychondritis-associated aortitis.

EULAR rheumatology open·2026

Same author

Five years of the Digital Rheumatology Network: insights and future directions.

EULAR rheumatology open·2026

Same journal

Intraloop neoangiogenesis in an AI-classified scleroderma pattern: recognizing a morphological clue to suspected dermatomyositis.

Diagnosis (Berlin, Germany)·2026

Same journal

Lessons in clinical reasoning: a case of a boy who continued crying after falling.

Diagnosis (Berlin, Germany)·2026

Same journal

Accuracy of emergency physicians' probability estimates for acute coronary syndrome.

Diagnosis (Berlin, Germany)·2026

Same journal

Interfering factors in the normative diagnostic approach.

Diagnosis (Berlin, Germany)·2026

Same journal

Using generative AI to support clinical reasoning coaching: a theory-informed approach.

Diagnosis (Berlin, Germany)·2026

Same journal

Learning from what went right: a Safety-II application of the SIDER protocol to a case of occult breast cancer.

Diagnosis (Berlin, Germany)·2026

查看所有相关文章

此摘要是机器生成的。

大型语言模型 (LLM) 显示出从图像中诊断常见皮肤疾病的潜力,GPT-4o实现了最高的准确性. 然而,性能各不相同,这凸显了需要谨慎使用和进一步开发可靠的临床决策支持的需要.

科学领域:

皮肤病学中的人工智能
医学图像分析医学图像分析
临床决策支持系统临床决策支持系统

背景情况:

解读皮肤学发现对公众和医疗专业人士来说都是一个挑战.
大型语言模型 (LLM) 提供了可访问的诊断支持的潜力,但它们在皮肤病学中的有效性尚未得到充分证实.

研究的目的:

用皮肤学图像评估各种多式联络LLM的诊断性能.
为了比较不同LLMs在识别常见皮肤疾病的准确性.

主要方法:

分析了50张皮肤学图像,其中包括牛皮,白风,疹和疹.
七个多模式LLM,包括GPT-4o,Gemini 1.5 Pro和Claude 3.5 Sonnet,使用标准化提示符进行了测试,用于顶级诊断生成.

主要成果:

GPT-4o获得了最高的整体精度 (67.8%),其次是GPT-4o mini (63.8%) 和Llama3.2 11B (61.4%).
诊断准确度在各病症之间有所不同,牛皮显示最高的平均LLM准确度 (59.2%) 和疹最低 (33.4%).
清晰的疾病特征提高了LLM准确性,而Llama3.2 90B则拒绝诊断私密区域的图像.

结论:

关键词:

聊天GPT 聊天GPT 聊天人工智能的人工智能是人工智能.皮肤学皮肤学诊断诊断诊断诊断诊断诊断大型语言模型.皮肤病理学皮肤病理学

更多相关视频

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

相关实验视频

Last Updated: Mar 1, 2026

Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application

Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application

Published on: April 14, 2023

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

皮肤学图像诊断中的LLM性能是可变的,需要仔细应用.
一个免费的,可在本地部署的LLM实现了大约66%的准确性,表明了安全,可访问的临床工具的潜力.
提高LLM准确性和整合临床数据可以增强诊断支持系统.