Scholarlink

Generative language models have changed the landscape of artificial intelligence in recent years. However, despite their advanced capabilities, they are prone to generate misleading results and may invent answers. In Spain, the National Health System has collected numerous health guides to inform medical procedures and protocols. In this paper, we utilize advanced generalist language models to extract relevant information and analyze validated content from health clinical guidelines. This approach offers innovative automated support for evidence-based clinical decision making. In our proposal, each of the system’s responses must track the source of the clinical evidence on which it is based, protecting users from hallucinatory responses. To study its feasibility in different medical settings, four clinical guidelines have been evaluated by human medical experts showing high reliability and traceability of evidence.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Leveraging Retrieval-Augmented Generation for Reliable Medical Question Answering Using Large Language Models

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Ksenia Kharitonova,
  • David Pérez-Fernández,
  • Javier Gutiérrez-Hernando,
  • Asier Gutiérrez-Fandiño,
  • Zoraida Callejas,
  • David Griol

摘要

Generative language models have changed the landscape of artificial intelligence in recent years. However, despite their advanced capabilities, they are prone to generate misleading results and may invent answers. In Spain, the National Health System has collected numerous health guides to inform medical procedures and protocols. In this paper, we utilize advanced generalist language models to extract relevant information and analyze validated content from health clinical guidelines. This approach offers innovative automated support for evidence-based clinical decision making. In our proposal, each of the system’s responses must track the source of the clinical evidence on which it is based, protecting users from hallucinatory responses. To study its feasibility in different medical settings, four clinical guidelines have been evaluated by human medical experts showing high reliability and traceability of evidence.