CLEAR: an auditable foundation model for radiology grounded in clinical concepts.

AI / informaticsNature biomedical engineering· Jul 2026

CLEAR: an auditable foundation model for radiology grounded in clinical concepts.

Tianyu Han, Riga Wu, Yu Tian … Daniel Truhn

Read paper

RadiologyAI / informaticsQ1IF 18.7

Auditable radiology AI model trained on nearly a million image-report pairs.

CLEAR: an auditable foundation model for radiology grounded in clinical concepts.

Tianyu Han … Daniel Truhn

Nature biomedical engineering · Jul 2026 · Q1 in General Medicine

Background

'Black box' deep learning models for medical image interpretation limit clinical trust and analysis of performance degradation.

Results

0.87

million

the AI model learned from nearly a million image-report pairs

n = 239,391 patients

More results

Here we introduce Concept-Level Embeddings for Auditable Radiology (Concept-Level Embeddings for Auditable Radiology), an auditable foundation model based on clinical concepts.

“

Conclusion

By integrating clinical knowledge directly into its reasoning process, Concept-Level Embeddings for Auditable Radiology offers a framework for robust model auditing, safer deployment and enhanced physician-AI collaboration, advancing towards trustworthy medical AI.

CLEAR: an auditable foundation model for radiology grounded in clinical concepts.

CLEAR: an auditable foundation model for radiology grounded in clinical concepts.

Related papers