24 January 2024 | Oscar A. Garcia Valencia, Charat Thongprayoon, Caroline C. Jadlowiec, Shennen A. Mao, Napat Leeaphorn, Pooja Budhiraja, Jasmina M. Craici, Maria L. Gonzalez Suarez, Wisit Cheungpasitporn
This study evaluates the capabilities of ChatGPT, specifically versions 3.5 and 4.0, in translating 54 English kidney transplant FAQs into Spanish. The translations were assessed for linguistic accuracy and cultural sensitivity by two native Spanish-speaking nephrologists of Mexican heritage using a 1-5 rating scale. The inter-rater reliability was 0.85, indicating high agreement. Both versions of ChatGPT achieved high levels of accuracy and cultural sensitivity, with GPT-4.0 slightly outperforming GPT-3.5 in accuracy. The study highlights the potential of AI-driven translations to improve health equity by enhancing access to essential kidney transplant information for Spanish-speaking Hispanic patients. However, the study also notes limitations, such as the narrow focus on a specific medical context and the reliance on human evaluators, suggesting the need for broader and more objective assessments in future research. The findings underscore the importance of AI in bridging language gaps and promoting health equity, particularly for linguistically diverse populations.This study evaluates the capabilities of ChatGPT, specifically versions 3.5 and 4.0, in translating 54 English kidney transplant FAQs into Spanish. The translations were assessed for linguistic accuracy and cultural sensitivity by two native Spanish-speaking nephrologists of Mexican heritage using a 1-5 rating scale. The inter-rater reliability was 0.85, indicating high agreement. Both versions of ChatGPT achieved high levels of accuracy and cultural sensitivity, with GPT-4.0 slightly outperforming GPT-3.5 in accuracy. The study highlights the potential of AI-driven translations to improve health equity by enhancing access to essential kidney transplant information for Spanish-speaking Hispanic patients. However, the study also notes limitations, such as the narrow focus on a specific medical context and the reliance on human evaluators, suggesting the need for broader and more objective assessments in future research. The findings underscore the importance of AI in bridging language gaps and promoting health equity, particularly for linguistically diverse populations.