Artificial Intelligence Improves Readability in Rheumatology Literature
Artificial intelligence can simplify complex rheumatology literature, making it easier for patients to understand essential health information.
Inadequate health literacy is a persistent challenge that affects outcomes for individuals with chronic diseases, including rheumatic conditions. Most peer-reviewed medical literature demands a college-level reading ability, far above the sixth-grade level recommended for public education materials. This study evaluated whether artificial intelligence, specifically ChatGPT 4.0, could bridge this gap by rewriting scientific rheumatology articles in simpler language.
Evaluating AI in Patient Education
Researchers analyzed twelve open-access rheumatology papers written by the senior study investigators. Using the Flesch-Kincaid Grade Level and Simple Measure of Gobbledygook indices, they assessed baseline readability and then re-evaluated the AI-generated summaries. Two rheumatology experts reviewed the simplified versions to verify accuracy and completeness.
The results were significant. ChatGPT reduced the average reading level from approximately the 15th to the 10th grade, a measurable improvement in accessibility (P<0.0001). The experts confirmed that the simplified summaries remained clinically accurate and appropriate. Word count was also substantially decreased, from an average of 3,517 words to 446 (P=0.047), indicating enhanced conciseness and focus.
Limitations and Clinical Implications
Despite these gains, the simplified texts still exceeded the ideal sixth-grade level for patient materials. This suggests that while large language models can advance health literacy, expert review remains essential to ensure accuracy and comprehension. For healthcare professionals, such tools may serve as an initial step toward generating readable, educational resources that better support informed patient decision-making in rheumatology care.
Reference: Mendoza-Pinto C et al. Artificial intelligence in patient education: evaluating large language models for understanding rheumatology literature. Front Digit Health. 2025;7:1623399.
