Key points are not available for this paper at this time.
Purpose: Patient information videos are excellent for conveying information on eye health. Language barriers lead to inaccessibility for ethnic minorities. So far, overcoming language barriers have been very expensive, but in this short communications paper, we share our experiences with an inexpensive generative artificial intelligence-based translation system for videos. Design: Explorative study. Methods: We developed a patient information video on a very common and broadly relevant issue: how to use eye drops. The original video was made in Danish. We used HeyGen (HeyGen, Los Angeles, California, USA) to translate the video into three categories according to distance from Danish according to comparative linguistics: highly related (English and German), remotely related (French and Polish), and no recognizable relationship (Arabic and Turkish). Ophthalmologists with high proficiency in Danish and each of these languages evaluated and commented on the accuracy of the translations. Results: All translations resulted in a recognizable clone of the original individual with synchronized lip movements and understandable language. We observed certain inaccuracies in the translation, however, these differed across languages without a specific pattern. Inconsistencies in formal/informal pronouns were observed across languages. But overall, the general information was conveyed across all languages. Conclusion: Modern generative artificial intelligence-based translation tools can help tearing down language barriers and improve accessibility of patient information videos in ophthalmology.
Eriksen et al. (Mon,) studied this question.