Improving Accuracy and Source Transparency in Responses to Soft Tissue Sarcoma Queries Using GPT-4o Enhanced with German Evidence-Based Guidelines
The RAG approach significantly enhanced the performance of GPT-4o in answering STS-related questions. However, the model still produced incorrect responses in certain complex scenarios. GPT-4o, even with RAG, should be used cautiously in clinical settings, particularly for rare diseases like sarcoma. Human expertise remains irreplaceable in medical decision-making.
Authors
- Franka Menge
- Cui Yang (ORCID: https://orcid.org/0000-0003-3779-0424)
- Christoph Reißfelder (ORCID: https://orcid.org/0000-0003-4591-1046)
- Peter Hohenberger (ORCID: https://orcid.org/0000-0001-5359-5923)
- Weiwei Jia (ORCID: https://orcid.org/0000-0001-7318-8997)
- Jens Jakob (ORCID: https://orcid.org/0000-0002-8499-7172)
- Chengpeng Li (ORCID: https://orcid.org/0000-0001-6751-2770)
- Yanjun Chu
- Tobias Speer
Institutions
- German Cancer Research Center (DE)
- Heidelberg University (DE)
- Ludwigshafen University of Business and Society (DE)
- University Hospital Heidelberg (DE)
- Ministry of Education (ET)
- University Medical Centre Mannheim (DE)
Publication Details
- Journal
- Oncology Research and Treatment
- Published
- 2025-02-28
- DOI
- https://doi.org/10.1159/000544978
- Citations
- 3
- Primary Topic
- Sarcoma Diagnosis and Treatment
- Type
- article
- Field-Weighted Citation Impact
- 4.56