Back
Top 0.2%
18.8%
Top 1%
9.5%
Top 2%
9.5%
Top 25%
8.3%
Top 1%
6.8%
Top 0.8%
6.8%
Top 15%
6.8%
Top 2%
6.3%
Top 2%
3.2%
Top 0.7%
3.2%
Top 91%
2.1%
Top 3%
2.1%
Top 8%
2.1%
Top 6%
1.0%
Top 8%
1.0%
Top 1%
1.0%
Top 6%
0.8%
Top 6%
0.8%
Top 16%
0.8%
Top 13%
0.8%
Top 6%
0.8%
Top 4%
0.8%
CLEAR: An Auditable Foundation Model for Radiology Grounded in Clinical Concepts
2026-01-17
health informatics
Title + abstract only
View on medRxiv
Show abstract
"Black box" deep learning models for medical image interpretation limit clinical trust and analysis of performance degradation. Here, we introduce Concept-Level Embeddings for Auditable Radiology (CLEAR), an auditable foundation model based on clinical concepts. Trained on over 0.87 million image-report pairs from 239,091 patients, CLEAR learns a visual representation and projects chest X-rays into a semantically rich space defined by large language model embeddings, making every prediction trac...
Predicted journal destinations
1
npj Digital Medicine
85 training papers
2
PLOS Digital Health
88 training papers
3
Journal of the American Medical Informatics Association
53 training papers
4
Scientific Reports
701 training papers
5
Journal of Biomedical Informatics
37 training papers
6
BMC Medical Informatics and Decision Making
36 training papers
7
Nature Communications
483 training papers
8
JAMIA Open
35 training papers
9
Computers in Biology and Medicine
39 training papers
10
JMIR Medical Informatics
16 training papers
11
PLOS ONE
1737 training papers
12
International Journal of Medical Informatics
25 training papers
13
Journal of Medical Internet Research
81 training papers
14
BMC Medical Research Methodology
41 training papers
15
eBioMedicine
82 training papers
16
Patterns
15 training papers
17
Diagnostics
36 training papers
18
Medicine
29 training papers
19
Nature Medicine
88 training papers
20
PLOS Computational Biology
141 training papers
21
Communications Medicine
63 training papers
22
Scientific Data
30 training papers