Performance of single-agent and multi-agent language models in Spanish language medical competency exams
Multi-agent strategies, particularly MDAGENTS, significantly enhance GPT-4o's performance on Spanish-language medical exams, leveraging collaboration to improve diagnostic accuracy. However, simpler single-agent strategies are sufficient to address many questions, high-lighting that only a fraction of standardized medical exams require sophisticated reasoning or multi-agent interaction. These findings suggest potential for LLMs as efficient and scalable tools in Spanish-speaking healthcare, though computational optimization remains a key area for future research.
Authors
- Ignacio Villagrán (ORCID: https://orcid.org/0000-0003-3130-8326)
- Fernando Altermatt (ORCID: https://orcid.org/0000-0002-0464-8643)
- Héctor J. Lacassie (ORCID: https://orcid.org/0000-0001-5758-4113)
- Marcelo Mendoza (ORCID: https://orcid.org/0000-0002-7969-6041)
- Andrés Neyem (ORCID: https://orcid.org/0000-0002-5734-722X)
- Nicolás I. Sumonte
Institutions
- Pontificia Universidad Católica de Chile (CL)
- Agencia Nacional de Investigación y Desarrollo (CL)
- Millennium Institute on Immunology and Immunotherapy (CL)
Publication Details
- Journal
- BMC Medical Education
- Published
- 2025-05-07
- DOI
- https://doi.org/10.1186/s12909-025-07250-3
- Citations
- 10
- Primary Topic
- Artificial Intelligence in Healthcare and Education
- Type
- article
- Field-Weighted Citation Impact
- 6.54
Funders
- Agencia Nacional de Investigación y Desarrollo
- Fondo de Fomento al Desarrollo Científico y Tecnológico