Assessing the Accuracy of Large Language Models on European Guidelines for Cervical Cancer: An In Silico Benchmarking Study
All models demonstrated suboptimal accuracy in aligning with clinical guidelines. ChatGPT 4.0 was the most accurate and consistent whereas DeepSeek R1 underperformed. Despite similar reliability across models, expert oversight remains essential to ensure safe clinical application and prevent misinformation.
Authors
- Matteo Pavone (ORCID: https://orcid.org/0000-0002-9791-9464)
- Jacques Marescaux
- Francesco Fanfani (ORCID: https://orcid.org/0000-0003-1991-7284)
- Nicolò Bizzarri (ORCID: https://orcid.org/0000-0002-1727-904X)
- Lise Lecointre (ORCID: https://orcid.org/0000-0002-1241-3094)
- Antonella Carcagnì
- Carlos Cantarini (ORCID: https://orcid.org/0000-0002-9372-9209)
- Barbara Costantini (ORCID: https://orcid.org/0000-0002-1414-0106)
- David Cibula (ORCID: https://orcid.org/0000-0001-6387-9356)
- Andrea Rosati (ORCID: https://orcid.org/0000-0003-3126-9400)
- Denis Querleu (ORCID: https://orcid.org/0000-0002-8240-8231)
- Anna Fagotti (ORCID: https://orcid.org/0000-0002-0465-5990)
- Chiara Innocenzi (ORCID: https://orcid.org/0009-0001-6997-9099)
- Massimo Criscione (ORCID: https://orcid.org/0000-0002-1627-2444)
- Nicola Macellari (ORCID: https://orcid.org/0009-0000-3547-4853)
Institutions
- Università Cattolica del Sacro Cuore (IT)
- Centre National de la Recherche Scientifique (FR)
- Charles University (CZ)
- Agostino Gemelli University Polyclinic (IT)
- Saint Camillus International University of Health and Medical Sciences (IT)
- Laboratoire des Sciences de l'Ingénieur, de l'Informatique et de l'Imagerie (FR)
- Institut de Recherche contre les Cancers de l’Appareil Digestif (FR)
- Institut de Chirurgie Guidée par l'Image (FR)
- General University Hospital in Prague (CZ)
- Université de Strasbourg (FR)
Publication Details
- Journal
- BJOG An International Journal of Obstetrics & Gynaecology
- Published
- 2025-11-24
- DOI
- https://doi.org/10.1111/1471-0528.70095
- Primary Topic
- Artificial Intelligence in Healthcare and Education
- Type
- article
- Field-Weighted Citation Impact
- 0.00