最先进的大型语言模型在ACR培训考试中的性能比较:2025年更新
Austin Young1, Rinald Paloka2, Ariba Islam3
1Northwell Health, Mather Hospital, Port Jefferson, New York (A.Y.).
Academic radiology
|September 25, 2025
概括
较新的大型语言模型 (LLM) 在放射学委员会考试中表现得更好. GPT-o1,GPT-4o和GPT-o3在准确度方面领先,这表明LLM可以在最小的数据污染问题下帮助居民学习.
科学领域:
- 医疗教育中的人工智能
- 放射学委员会考试的准备工作 放射学委员会考试的准备工作
- 大型语言模型 (LLM) 绩效评估
背景情况:
- 之前的研究评估了大型语言模型 (LLM) 在放射学委员会风格评估中的表现.
- 这项研究扩展了之前的工作,通过评估ACR诊断放射学培训考试 (DXIT) 的新型LLM.
研究的目的:
- 评估尖端LLMs (GPT-4o,GPT-o1,GPT-o3,Claude,Gemini,Grok) 在标准化DXIT问题上的表现.
- 为了比较基于文本和基于图像的问题的模型准确性,以评估多模式推理.
- 通过比较原始问题与修订问题的表现来调查潜在数据污染的影响.
主要方法:
- 通过使用106个公开可用的DXIT问题,评估了7个LLM.
- 模型使用标准化的指令集被提示模拟居民的反应.
- 计算了未调整和逻辑调整的准确性,并对基于文本和图像的问题进行了子组分析. 用修订的问题来测试数据污染.
主要成果:
- GPT-o1 (71.7%),GPT-4o (69.8%) 和GPT-o3 (68.9%) 实现了最高的未经调整的精度.
- 在逻辑调整精度方面也观察到类似的趋势,GPT-o1,GPT-4o和GPT-o3的表现优于其他模型.
- 在基于文本的问题上,GPT-4o的表现明显更好;在修订后的问题上的表现与原始问题相比,这表明数据污染有限.
结论:
- 现代的LLM,特别是来自OpenAI的LLM,在放射学董事会式评估方面表现强并有所改善.
- 对修订后提示的可比性能表明数据污染的作用有限.
- 士课程显示出通过个性化的反和实践来支持放射学住院教育的巨大潜力.
相关概念视频
Improving Translational Accuracy
3.5K
3.5K
Improving Translational Accuracy
14.1K
Base complementarity between the three base pairs of mRNA codon and the tRNA anticodon is not a failsafe mechanism. Inaccuracies can range from a single mismatch to no correct base pairing at all. The free energy difference between the correct and nearly correct base pairs can be as small as 3 kcal/ mol. With complementarity being the only proofreading step, the estimated error frequency would be one wrong amino acid in every 100 amino acids incorporated. However, error frequencies observed in...
14.1K
Multiple Comparison Tests
4.4K
Multiple comparison test, abbreviated as MCT, is a post hoc analysis generally performed after comparing multiple samples with one or more tests. An MCT will help identify a significantly different sample among multiple samples or a factor among multiple factors.
It would be easy to compare two samples using a significance alpha level of 0.05. In other words, there is only one sample pair to be compared. However, it would be difficult to identify a significantly different sample if the number...
It would be easy to compare two samples using a significance alpha level of 0.05. In other words, there is only one sample pair to be compared. However, it would be difficult to identify a significantly different sample if the number...
4.4K
