Large Language Models May Help Patients Understand Peer-Reviewed Scientific Articles About Ophthalmology: Development...

Area of Science:

• Ophthalmology
• Medical Informatics
• Health Literacy

Background:

• Health literacy is crucial for public health, with a recommended reading level of 6th grade for patient information.
• Scientific articles are often too complex for the general population to understand.
• Previous work explored large language models (LLMs) for patient-targeted health information.

Purpose of the Study:

• To investigate the use of LLMs, specifically ChatGPT, to improve the readability of ophthalmology research.
• To assess the effectiveness of LLMs in making peer-reviewed scientific literature more accessible.

Main Methods:

• Selected 12 open-access, peer-reviewed ophthalmology papers.
• Assessed readability using Flesch-Kincaid Grade Level and Simple Measure of Gobbledygook.
• Used ChatGPT 4.0 to simplify the text to a 6th-grade reading level, with evaluation by uveitis specialists.

Main Results:

• ChatGPT significantly reduced the reading level of papers from 15th to 7th grade (P<.001).
• The simplified text was deemed appropriate by expert ophthalmologists.
• The length of the articles was also reduced.

Conclusions:

• LLMs show potential for enhancing health literacy.
• LLMs can make complex medical literature more accessible to the general public.
• Direct interaction with scientific articles is facilitated by LLM-driven simplification.