AUTHOR=Wu Gloria , Lee David A. , Zhao Weichen , Wong Adrial , Sidhu Sahej 

TITLE=ChatGPT: is it good for our glaucoma patients?

JOURNAL=Frontiers in Ophthalmology

VOLUME=Volume 3 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/ophthalmology/articles/10.3389/fopht.2023.1260415

DOI=10.3389/fopht.2023.1260415

ISSN=2674-0826

ABSTRACT=Purpose: Our study investigates ChatGPT and its ability to communicate with glaucoma patients. Methods: We posed eight questions to ChatGPT and used The Flesch-Kincaid, Gunning Fog, SMOG Index, and Dale-Chall Readability Formulas to evaluate its accessibility for patients. We compared its answers to those found in the American Academy of Ophthalmology website's "Glaucoma" section.Results: ChatGPT's texts required reading comprehension of a higher grade level (average=12.5 grade + 1.6) than that of AAO.org (average=9.4 grade + 3.5), (0.0384). For the eight responses, ChatGPT used the ophthalmic key terms 34/86 times vs 86/86 times in AAO.org. ChatGPT mentioned "eye doctor" once, never using the formal term, Ophthalmologist. The term ophthalmologist has been used 26 times by the AAO website. The word counts of ChatGPT and AAO.org's answers were similar (p=0.571), with phrases of a homogenous length. Conclusion: ChatGPT trains on texts, phrases, and algorithms inputted by software engineers. As ophthalmologists, through our websites and journals, we should consider encoding the phrase "see an ophthalmologist." Our medical assistants should sit with patients to ensure the correct dissemination of information. ChatGPT is effective for providing general information such as definitions or potential treatment options for 2 glaucoma. However, ChatGPT has a proclivity towards repetitive answers that could be too difficult for a patient to read due to its elevated readability scores.ChatGPT trains on texts, phrases, and algorithms inputted by patients all around the globe. As ophthalmologists, through our websites and journals, we should consider encoding the phrase "see an ophthalmologist" as well as considering hiring medical assistants or diabetes educators to relay the effective and correct information to our patients, which they may not receive through artificial intelligence. This type of artificial intelligence, like ChatGPT, may be effective for giving out generic information such as definitions or potential treatments for glaucoma but has a proclivity towards giving out repetitive or ineffectual answers that could potentially be too difficult for a patient to read due to its elevated readability scores.