相关实验视频

Updated: Jun 27, 2026

Guidelines and Experience Using Imaging Biomarker Explorer IBEX for Radiomics

Guidelines and Experience Using Imaging Biomarker Explorer IBEX for Radiomics

Published on: January 8, 2018

使用大型语言模型评估放射学研究的方法质量:METRICS-E3框架的附加值

Ismail Mese¹, Burak Kocak²

¹Department of Radiology, Uskudar State Hospital, Istanbul, Turkey.

European journal of radiology

|November 22, 2025

概括

METRICS-E3资源增强了大型语言模型 (LLM) 评估放射学研究质量的能力,改善了与人类专家的协议. 这种LLM辅助的评估显示了作为预先选和审计的可扩展工具的承诺.

科学领域:

医学成像和人工智能医学成像和人工智能
放射学研究方法论研究方法论
科学中的自然语言处理.

背景情况:

评估放射学研究的方法质量对于可靠的临床翻译至关重要.
大型语言模型 (LLM) 显示了自动化研究评估的潜力,但其准确性需要改进.
方法论辐射ICS评分 (METRICS) 是评估辐射学研究质量的工具.

研究的目的:

评估METRICS-E3 (用示例解释和阐述) 资源是否可以提高LLM在使用METRICS框架评估放射学研究质量的表现.
为了比较LLM与METRICS-E3的和没有METRICS-E3的LLM绩效与人类共识和其他LLM基准.

主要方法:

一项元研究研究评估了48篇放射学文章,使用METRICS工具与两个GPT-5管道进行评估:基线 (GPT-5) 和用METRICS-E3 (GPT-5 E3) 增强.
每篇文章在每个管道中都被评估了三次,结果通过多数投票汇总.
将GPT-5和GPT-5E3的结果与之前报告的GPT-4o数据和参考人类共识进行了比较.

主要成果:

虽然GPT-4o获得了最高的METRICS中位数 (79.50%),但与基线GPT-5相比,GPT-5 E3与人类评级的一致性有所改善 (肯德尔的 τ从0.474增加到0.626).
通过类内相关系数衡量的协议从0.539 (GPT-5) 增加到0.793 (GPT-5 E3).
在GPT-5 E3中,通过METRICS.评估的项目中有83.3%的项目和80%的条件得到了更好的同意.

结论:

METRICS-E3资源显著改善了GPT-5与人类共识的调整以及评估放射学方法质量的可靠性.
使用METRICS和METRICS-E3的LLM辅助评估可以作为在专家监督下进行预选或审计的可扩展辅助.
为了证实这些发现,需要对各种数据集,人类读者和LLM架构进行进一步的验证.

关键词:

人工智能的人工智能是人工智能.准则准则准则准则大型语言模型.测量指标 (METRICS) 是一个指标.METRICS-E3 测量仪-E3 测量仪-E3 测量仪-E3 测量仪-E3 测量仪机器学习是机器学习.无线电学 (Radiomics) 是一种无线电学.质感分析质感分析

更多相关视频

Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model

Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model

Published on: August 16, 2020

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

相关概念视频

Methods to Assess Microbial Communities

Methods to Assess Microbial Communities

Microbial communities, comprising bacteria, archaea, and eukaryotic microorganisms, inhabit diverse ecosystems and play crucial roles in environmental and biological processes. Their diversity is defined by three main parameters: species richness (the number of distinct species), species abundance (the relative quantity of each species), and species evenness (how uniformly individual species are distributed in various locations). These factors together shape the structure and ecological balance...

Methods of Medium Optimization

Methods of Medium Optimization

Optimizing growth media enhances microbial proliferation and maximizes product yield. Statistical experimental design methodologies provide structured and reproducible approaches, offering progressively higher levels of robustness and efficiency.The One-Factor-at-a-Time (OFAT) MethodThe One-Factor-at-a-Time (OFAT) method involves adjusting a single variable while keeping all others constant. However, it cannot detect interactions between variables, often leading to suboptimal outcomes when...