Abstract
Purpose :
Glaucoma is the leading cause of blindness in Hispanics over the age of 40. Social factors, such as lower patient education and literacy levels, are known factors related to worse health outcomes for this population. Many studies have demonstrated success in utilizing artificial intelligence (AI) to improve the readability of patient education materials (PEMs) in English. Our study is looking at the ability of AI to improve the readability of PEMs in the Spanish language.
Methods :
A Google search was conducted in Spanish with the following phrases: “glaucoma”, “glaucoma qué es”, “glaucoma síntomas”, and “glaucoma tratamiento”. The first 10 search results that appeared when searching each of the phrases were evaluated for inclusion in the study. Exclusion criteria consisted of any PEMs part of a sponsored advertisement or fully written in English. PEMs that met the criteria were input into an AI program, ChatGPT, to be re-worded to a 6th-grade reading level. Índice Flesch-Szigriszt (INFLESZ) score and percent complex words of the PEMs were calculated before and after ChatGPT input. INFLESZ score ranges corresponded to a specific level of difficulty (Table 1). A paired sample two-tailed T-test was used for analysis.
Results :
17 PEMs met the criteria with 47% having an initial INFLESZ score in the normal range (n=8). The search phrases “glaucoma que es” and “glaucoma tratamiento” yielded 70% of the included articles (n=12). The average INFLESZ score of the original PEMs was 59.95 ± 8.68 corresponding to a “normal” score. On average the original PEMs contained about 30.45% complex words per article. Post ChatGPT PEM INFLESZ score was found to be significantly higher (p<0.05) with an average of 66.20 ± 7.29 corresponding to a “quite easy” score and 28.15% complex words per article. There was no significant difference in the percentage of complex words. Change in INFLESZ score by article is presented in Table 2.
Conclusions :
AI may be a useful tool in improving the readability of glaucoma PEMs for Spanish-speaking patients. Additionally, patients with lower health literacy levels may be able to utilize ChatGPT as a resource to understand complicated information regarding their glaucoma diagnosis and treatment. Future studies should expand on whether the AI-generated articles are conveying important health information with the same urgency as in the original articles.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.