Can open source large language models be used for tumor documentation in Germany? -- An evaluation on urological doctors' notes
2025-08-08
Summary
The article evaluates the potential of open-source large language models (LLMs) for automating tumor documentation in Germany, focusing on urological doctors' notes. It assesses 11 different LLMs on their ability to identify tumor diagnoses, assign ICD-10 codes, and extract the date of diagnosis using a specially prepared dataset of anonymized notes.
Why This Matters
This research is significant as it addresses the current manual and labor-intensive process of tumor documentation in Germany, which is critical for improving oncological care and research. Automating this process with LLMs could enhance efficiency, reduce errors, and free up medical staff for more critical tasks.
How You Can Use This Info
For healthcare professionals and administrators, understanding the capabilities of LLMs can inform decisions about integrating technology into clinical documentation workflows. Organizations can explore using open-source LLMs, especially those with 7-12 billion parameters, as practical tools for improving documentation accuracy and efficiency in cancer care. Additionally, resources like the dataset and evaluation code provided in the study can serve as a foundation for further development and customization in their specific contexts.