RadTranslateGPT: An Improved AI-Based System for Translation and Simplification of Structured Radiology Reports.
Authors
Affiliations (4)
Affiliations (4)
- Medically Engineered Solutions in Healthcare Incubator, Innovation in Operations Research Center (MESH IO), Mass General Brigham, Boston, MA; University of Missouri, Kansas City School of Medicine.
- Medically Engineered Solutions in Healthcare Incubator, Innovation in Operations Research Center (MESH IO), Mass General Brigham, Boston, MA.
- Harvard Medical School, Boston, MA; Medically Engineered Solutions in Healthcare Incubator, Innovation in Operations Research Center (MESH IO), Mass General Brigham, Boston, MA; Department of Radiology, Mass General Brigham, Boston, MA.
- Harvard Medical School, Boston, MA; Medically Engineered Solutions in Healthcare Incubator, Innovation in Operations Research Center (MESH IO), Mass General Brigham, Boston, MA; Department of Radiology, Mass General Brigham, Boston, MA. Electronic address: [email protected].
Abstract
Radiology reports often contain complex medical jargon that can be difficult for patients to understand, especially those with limited English proficiency (LEP). With increased access to electronic health records, there is a need for patient-friendly and multilingual interpretation of radiology reports. This study evaluated the performance of GPT-o1 in simplifying and translating emergency radiology reports into Spanish, Arabic, and Mandarin, and compared it with Google Translate. Thirty de-identified emergency radiology reports were selected from an institutional database. Phase 1 involved the evaluation of the GPT-o1 simplified reports by three board-certified emergency radiologists. Phase 2 involved nine medical interpreters (three per language) assessing both GPT-o1 and Google Translate outputs. Both groups rated outputs on a 5-point Likert scale. There were a total of ninety evaluations of the simplified radiology reports. 76.9% (n=69) of the reports were rated as "Extremely accurate" or "Very accurate," 45.6% (n=41) were "Very Clear," and 98.9% (n=89) were marked as "useful" for patient comprehension. GPT-o1 achieved significantly higher translational accuracy (median 4.0 vs 3.0; p < .001), higher language register scores, and was rated more comprehensible than Google Translate in all languages tested. GPT-o1 generated simplified, patient-friendly radiology reports and also produced high-quality translations into the three languages tested; however, these findings should be interpreted in the context of this pilot study with limited sample size.These findings suggest that LLMs could serve as tools to enhance health literacy for LEP populations. Further validation of these LLMs is needed before clinical integration.