Scholarlink

Background <p>Language barriers in pediatric emergency medicine discharge instructions can impact patient safety, leading to poorer post-discharge outcomes. Google Translate has been widely used clinically as an alternative to professional translations but is prone to errors. ChatGPT-4 is a large language model that has been shown to offer improved translation abilities in various domains. This study compares ChatGPT-4 and Google Translate for translating discharge instructions into Arabic and Simplified Chinese.</p> Methods <p>15 standardized pediatric emergency department discharge instructions from our institution were translated from English to both Simplified Chinese and Arabic, using both ChatGPT-4 and Google Translate. Eight bilingual medical professionals (four per language group) were recruited as participants and presented with the English and the translated discharge instructions in a blinded manner. Participants rated the translations on a 5-point Likert scale for accuracy, safety, and understandability, followed by their overall translation preference.</p> Results <p>In Arabic, ChatGPT-4 was found to be more accurate than Google Translate, (mean 4.1 vs. 3.7, mean difference = 0.43, 95% CI 0.15, 0.72, <i>p</i> = 0.003). Similarly for Simplified Chinese, ChatGPT-4 was found to be more accurate than Google Translate (mean 4.4 vs. 4.2; mean difference = 0.28, 95% CI 0.07, 0.50, <i>p</i> = 0.010). There were no significant differences in safety or understandability between ChatGPT-4 and Google Translate in either language. ChatGPT-4-generated translations were consistently preferred over Google Translate in both languages.</p> Conclusion <p>Overall, ChatGPT-4 is a more accurate alternative to Google Translate in translating written pediatric discharge instructions from English to Simplified Chinese and Arabic. ChatGPT-4 could be a particularly valuable translation tool in healthcare settings with limited access to translation resources such as professional interpreters.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

ChatGPT-guided translation of pediatric emergency department discharge instructions

  • 期刊论文

  • 发表日期：

    2026年1月14日

查看全文
  • Andrew Yuanbo Wang,
  • Kristine Van Aarsen,
  • Sheena Belisle

摘要

Background

Language barriers in pediatric emergency medicine discharge instructions can impact patient safety, leading to poorer post-discharge outcomes. Google Translate has been widely used clinically as an alternative to professional translations but is prone to errors. ChatGPT-4 is a large language model that has been shown to offer improved translation abilities in various domains. This study compares ChatGPT-4 and Google Translate for translating discharge instructions into Arabic and Simplified Chinese.

Methods

15 standardized pediatric emergency department discharge instructions from our institution were translated from English to both Simplified Chinese and Arabic, using both ChatGPT-4 and Google Translate. Eight bilingual medical professionals (four per language group) were recruited as participants and presented with the English and the translated discharge instructions in a blinded manner. Participants rated the translations on a 5-point Likert scale for accuracy, safety, and understandability, followed by their overall translation preference.

Results

In Arabic, ChatGPT-4 was found to be more accurate than Google Translate, (mean 4.1 vs. 3.7, mean difference = 0.43, 95% CI 0.15, 0.72, p = 0.003). Similarly for Simplified Chinese, ChatGPT-4 was found to be more accurate than Google Translate (mean 4.4 vs. 4.2; mean difference = 0.28, 95% CI 0.07, 0.50, p = 0.010). There were no significant differences in safety or understandability between ChatGPT-4 and Google Translate in either language. ChatGPT-4-generated translations were consistently preferred over Google Translate in both languages.

Conclusion

Overall, ChatGPT-4 is a more accurate alternative to Google Translate in translating written pediatric discharge instructions from English to Simplified Chinese and Arabic. ChatGPT-4 could be a particularly valuable translation tool in healthcare settings with limited access to translation resources such as professional interpreters.