In this study, the authors compared the performance of three popular large learning models (LLMs) – ChatGPT3.5, ChatGPT4 and Gemini – in rewriting existent patient education materials on retinopathy of prematurity (RoP) to be more readable and accessible to patients, and to compare their ability to generate new patient education materials on RoP. Readability was evaluated using SMOG and Flesch-Kincaid grade level scores. Patient education materials generated by ChatGPT4 were best aligned to 6th grade reading levels. Online sources of patient education materials were of 9th grade level and, when rewritten by LLMs, all improved to 6th grade level. Despite generated materials being of high quality and readability, none were found to be actionable and this was likely due to the lack of specific action checklists for patients and visual aids. The authors conclude that LLMs can be useful as a supplementary tool to help automate the process of generating patient educational materials.
Using AI to generate and revise patient educational materials for retinopathy of prematurity
for retinopathy of prematurity
Implementing generative AI to enhance patient education of retinopathy of prematurity.
CONTRIBUTOR
Fiona Rowe (Prof)
Institute of Population Health, University of Liverpool, UK.
View Full Profile
