This study evaluates how open source large language models could transform tumor documentation workflows in German healthcare settings.

• Tested 11 different open source LLMs (1-70B parameters) on critical documentation tasks
• Assessed performance on tumor diagnosis identification, ICD-10 coding, and date extraction
• Evaluated potential for reducing manual documentation burden in oncology

This research addresses a significant healthcare efficiency challenge by exploring how AI can augment medical staff in structured data extraction from clinical notes, potentially improving documentation accuracy and reducing administrative workload.

Can open source large language models be used for tumor documentation in Germany?