Scholarlink

Background <p>Large language models (LLM) can automatically process clinical free-text documents, extract key information, and thereby reduce reading effort and documentation-related workload. High-quality data and targeted model control are essential for practical applicability.</p> Material and methods <p>Various approaches to information extraction are presented. Additionally, 24&#xa0;unstructured pathological reports of bone and soft tissue tumors are processed using the local, generic LLM Llama 4&#xa0;Scout with three different prompt variants and compared in terms of extraction quality.</p> Results <p>Prompt design had a&#xa0;substantial impact on model behavior. Prompts with clear parameter definitions and examples achieved the most reliable results. Typical LLM-specific errors, such as hallucinations and misclassifications, were also observed.</p> Summary <p>LLM can support clinical staff by rapidly and systematically extracting relevant content from free-text documents. Safe and effective use requires high-quality data, precise inputs, and close collaboration between medical and technical experts.</p>

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

LLM-gestützte Extraktion klinischer Daten: Potenziale und Herausforderungen

  • 期刊论文

  • 发表日期：

    2025年12月19日

查看全文
  • Paulina Seidl,
  • Marton Szep,
  • Sebastian Breden,
  • Fiona Charitou,
  • Carolin Mogler,
  • Peter Schüffler,
  • Rüdiger von Eisenhart-Rothe,
  • Igor Lazic,
  • Florian Hinterwimmer

摘要

Background

Large language models (LLM) can automatically process clinical free-text documents, extract key information, and thereby reduce reading effort and documentation-related workload. High-quality data and targeted model control are essential for practical applicability.

Material and methods

Various approaches to information extraction are presented. Additionally, 24 unstructured pathological reports of bone and soft tissue tumors are processed using the local, generic LLM Llama 4 Scout with three different prompt variants and compared in terms of extraction quality.

Results

Prompt design had a substantial impact on model behavior. Prompts with clear parameter definitions and examples achieved the most reliable results. Typical LLM-specific errors, such as hallucinations and misclassifications, were also observed.

Summary

LLM can support clinical staff by rapidly and systematically extracting relevant content from free-text documents. Safe and effective use requires high-quality data, precise inputs, and close collaboration between medical and technical experts.