Robustness tests for biomedical foundation models should tailor to specifications
The rise of biomedical foundation models creates new hurdles in model testing and authorization, given their broad capabilities and susceptibility to complex distribution shifts. We suggest tailoring robustness tests according to task-dependent priorities and propose to integrate granular notions of robustness in a predefined specification to guide implementation. Our approach facilitates the standardization of robustness assessments in the model lifecycle and connects abstract AI regulatory frameworks with concrete testing procedures.
Authors
- Madhumita Sushil (ORCID: https://orcid.org/0000-0001-7884-0526)
- Noah Baker (ORCID: https://orcid.org/0000-0002-6368-2083)
- Tom David (ORCID: https://orcid.org/0000-0003-3063-8657)
- Stefan Bauer (ORCID: https://orcid.org/0000-0003-1712-060X)
- Qiming Cui (ORCID: https://orcid.org/0009-0001-7993-3561)
- A Jay Holmgren (ORCID: https://orcid.org/0000-0002-7939-6831)
- Reza Abbasi-Asl (ORCID: https://orcid.org/0000-0001-7824-4628)
- R. Patrick Xian (ORCID: https://orcid.org/0000-0001-9895-6956)
Institutions
- University of California, San Francisco (US)
- Laboratoire Pluridisciplinaire de Recherche en Ingénierie des Systèmes, Mécanique et Energétique (FR)
- Isotopen Technologien München (Germany) (DE)
- Technical University of Munich (DE)
- University of California, Berkeley (US)
Publication Details
- Journal
- npj Digital Medicine
- Published
- 2025-08-29
- DOI
- https://doi.org/10.1038/s41746-025-01926-2
- Citations
- 3
- Primary Topic
- Software Testing and Debugging Techniques
- Type
- article
- Field-Weighted Citation Impact
- 6.92
Funders
- National Institutes of Health
- National Cancer Institute