今日神经病学最高分文献 - 2025-10-30

今日神经病学最高分文献 - 2025-10-30

From 萌喵读文献-神经病学 by Meng Zhao

October 30, 2025 · 4 min

About this episode

This episode discusses a significant study on the impact of web retrieval on the accuracy of large language models in neurology.

科研喵使用AI读文献，祝你效率百倍，访问labcat.com.cn下载。本期关注发表在影响因子5.8的《Journal of Medical Internet Research》上的研究"Evaluating Web Retrieval-Assisted Large Language Models With and Without Whitelisting for Evidence-Based Neurology: Comparative Study"。这项神经学领域的重要研究发现，当大型语言模型的网络检索限制在权威神经学网站时，回答准确率显著提升—Sonar模型从60%跃升至78%，专业版模型更达到88-89%。研究证实，使用专业医学来源可使模型表现媲美专业文献检索系统，为临床决策提供可靠支持。这一发现为AI医疗应用的安全边界提供了重要参考，表明简单的内容过滤就能显著提升AI在循证医学中的实用价值。

People in this episode

Host: Meng Zhao

Topics covered

  • neuroscience
  • AI in medicine
  • evidence-based neurology
  • large language models
  • clinical decision support

Keywords

  • neuroscience
  • AI
  • large language models
  • evidence-based medicine
  • clinical decision support
  • Journal of Medical Internet Research
  • Sonar model

Mentioned in this episode

Organizations: Journal of Medical Internet Research, labcat.com.cn

Books & works: Evaluating Web Retrieval-Assisted Large Language Models With and Without Whitelisting for Evidence-Based Neurology: Comparative Study

More episodes of 萌喵读文献-神经病学

Explore listener stats, chart rankings, contacts and more on the 萌喵读文献-神经病学 podcast page.