Multimodal BEHRT: transformers for multimodal electronic health records to predict breast cancer prognosis
Our work highlights both the potential of EHR data for improving our understanding of breast cancer and the ability of transformer-based architectures to learn from EHR data containing much fewer than the millions of records typically used in currently published studies. The representation of patient trajectories used by M-BEHRT captures their sequential aspect, and opens new research avenues for understanding complex diseases and improving patient care.
Authors
- Anne‐Sophie Hamy (ORCID: https://orcid.org/0000-0003-4430-3022)
- Michal Rosen‐Zvi (ORCID: https://orcid.org/0000-0001-7616-9724)
- Fabien Reyal (ORCID: https://orcid.org/0000-0002-2318-3589)
- Aullène Toussaint
- Élise Dumas (ORCID: https://orcid.org/0000-0002-7877-719X)
- Michael M. Danziger (ORCID: https://orcid.org/0000-0002-2674-0109)
- Julien Guérin (ORCID: https://orcid.org/0000-0002-3024-8483)
- Ndèye Maguette Mbaye
- Chloé‐Agathe Azencott (ORCID: https://orcid.org/0000-0003-1003-301X)
Institutions
- Inserm (FR)
- Hebrew University of Jerusalem (IL)
- Université Paris Sciences et Lettres (FR)
- Immunité et Cancer (FR)
- Institut Jean Godinot (FR)
- IBM Research - Haifa (IL)
- Institut Curie (FR)
Publication Details
- Journal
- Frontiers in Oncology
- Published
- 2025-10-17
- DOI
- https://doi.org/10.3389/fonc.2025.1496215
- Primary Topic
- Machine Learning in Healthcare
- Type
- article
- Field-Weighted Citation Impact
- 0.00