View in EDS

Privacy-ensuring Open-weights Large Language Models Are Competitive with Closed-weights GPT-4o in Extracting Chest Radiography Findings from Free-Text Reports.

Saved in:
Bibliographic Details
Title: Privacy-ensuring Open-weights Large Language Models Are Competitive with Closed-weights GPT-4o in Extracting Chest Radiography Findings from Free-Text Reports.
Authors: Nowak S; From the Department of Diagnostic and Interventional Radiology, University Hospital Bonn, Venusberg-Campus 1, 53127 Bonn, Germany., Wulff B; From the Department of Diagnostic and Interventional Radiology, University Hospital Bonn, Venusberg-Campus 1, 53127 Bonn, Germany., Layer YC; From the Department of Diagnostic and Interventional Radiology, University Hospital Bonn, Venusberg-Campus 1, 53127 Bonn, Germany., Theis M; From the Department of Diagnostic and Interventional Radiology, University Hospital Bonn, Venusberg-Campus 1, 53127 Bonn, Germany., Isaak A; From the Department of Diagnostic and Interventional Radiology, University Hospital Bonn, Venusberg-Campus 1, 53127 Bonn, Germany., Salam B; From the Department of Diagnostic and Interventional Radiology, University Hospital Bonn, Venusberg-Campus 1, 53127 Bonn, Germany., Block W; From the Department of Diagnostic and Interventional Radiology, University Hospital Bonn, Venusberg-Campus 1, 53127 Bonn, Germany., Kuetting D; From the Department of Diagnostic and Interventional Radiology, University Hospital Bonn, Venusberg-Campus 1, 53127 Bonn, Germany., Pieper CC; From the Department of Diagnostic and Interventional Radiology, University Hospital Bonn, Venusberg-Campus 1, 53127 Bonn, Germany., Luetkens JA; From the Department of Diagnostic and Interventional Radiology, University Hospital Bonn, Venusberg-Campus 1, 53127 Bonn, Germany., Attenberger U; From the Department of Diagnostic and Interventional Radiology, University Hospital Bonn, Venusberg-Campus 1, 53127 Bonn, Germany., Sprinkart AM; From the Department of Diagnostic and Interventional Radiology, University Hospital Bonn, Venusberg-Campus 1, 53127 Bonn, Germany.
Source: Radiology [Radiology] 2025 Jan; Vol. 314 (1), pp. e240895.
Publication Type: Journal Article; Research Support, Non-U.S. Gov't
Journal Info: Publisher: Radiological Society of North America Country of Publication: United States NLM ID: 0401260 Publication Model: Print Cited Medium: Internet ISSN: 1527-1315 (Electronic) Linking ISSN: 00338419 NLM ISO Abbreviation: Radiology Subsets: MEDLINE
Database: MEDLINE Ultimate
Description
ISSN:1527-1315
DOI:10.1148/radiol.240895