This evaluation aims to compare the accuracy of machine translation tools with professional interpreters for emergency department discharge instructions.
Conducted a multi-language evaluation of emergency department discharge instructions.
Compared the translations from Google Translate and ChatGPT-4o with those of professional interpreters.
Assessed translation accuracy across different domains.
Google Translate and ChatGPT-4o were found to be non-inferior to professional interpreters.
Most domains of translation accuracy showed similar results between machine tools and human interpreters.
Abstract
In this multi-language evaluation of real-world ED discharge instructions, Google Translate and ChatGPT-4o were non-inferior to professional interpreters for most domains of translation accuracy.