Health AI Central
Share:
Peer-Reviewed Publication
Nat Commun2026;17(1)June 11, 2026Journal Article

Towards robust foundation models for digital pathology.

Jonah Kömen1,2, Edwin D de Jong3, Julius Hense1,2, Hannah Marienwald1,2, Jonas Dippel1,2,4, Philip Naumann1,2, Eric Marcus5, Lukas Ruff4, Maximilian Alber4,6, Jonas Teuwen5, Frederick Klauschen7,8,9,10,11, Klaus-Robert Müller12,13,14,15
1Berlin Institute for the Foundations of Learning and Data (BIFOLD), Berlin, Germany.
2Machine Learning Group, Technische Universität Berlin, Berlin, Germany.
3Aignostics GmbH, Berlin, Germany. jong.de.edwin.work@gmail.com.
4Aignostics GmbH, Berlin, Germany.
5The Netherlands Cancer Institute Amsterdam (NKI), Antoni van Leeuwenhoek Hospital (AvL), Amsterdam, Netherlands.
6Institute of Pathology, Charité Universitätsmedizin Berlin, Berlin, Germany.
7Berlin Institute for the Foundations of Learning and Data (BIFOLD), Berlin, Germany. f.klauschen@lmu.de.
8Institute of Pathology, Charité Universitätsmedizin Berlin, Berlin, Germany. f.klauschen@lmu.de.
9Institute of Pathology, Ludwig-Maximilians-Universität München, Munich, Germany. f.klauschen@lmu.de.
10German Cancer Research Center, Heidelberg, and German Cancer Consortium, Munich, Germany. f.klauschen@lmu.de.
11Bavarian Center for Cancer Research (BZKF), Munich, Germany. f.klauschen@lmu.de.
12Berlin Institute for the Foundations of Learning and Data (BIFOLD), Berlin, Germany. klaus-robert.mueller@tu-berlin.de.
13Machine Learning Group, Technische Universität Berlin, Berlin, Germany. klaus-robert.mueller@tu-berlin.de.
14Department of Artificial Intelligence, Korea University, Seoul, Korea. klaus-robert.mueller@tu-berlin.de.
15Max-Planck Institute for Informatics, Saarbrücken, Germany. klaus-robert.mueller@tu-berlin.de.
PubMed PMC DOI PDF

Abstract

Biomedical Foundation Models (FMs) are transforming AI-enabled healthcare research and entering clinical validation. However, their susceptibility to learning non-biological features - including variations in laboratory procedures and scanner hardware - poses risks for clinical deployment. We introduce PathoROB, a public benchmark quantifying FM robustness to non-biological features. Representatio…

Create a free account to keep reading

Free members get 10 full research views every month across publications, clinical trials, FDA clearances, adverse events, and NIH grants. No credit card required.

Create Free AccountSign In

Want unlimited research access? See Pro plans

Data Accuracy Notice: Research intelligence on Health AI Central is aggregated from public sources (PubMed, ClinicalTrials.gov, FDA, NIH, CMS, and others) and refreshed nightly. Classifications and derived metrics are produced by automated methods described in our Methodology. We recommend verifying critical data points against the primary sources before making decisions.