Abstract 18238: ChatGPT Has the Potential to Increase Accessibility to Cardiology Research
Aorzala Atmar, Saman Andalib, Aidin Spina, John Dylewski, Tessa Haspil-Corgan- Physiology (medical)
- Cardiology and Cardiovascular Medicine
Introduction: Generative language models like ChatGPT have experienced a novel rise in their implementation. This study explored ChatGPT’s potential in simplifying cardiology research abstracts published in specific journals within the family of Journals of American College of Medicine (JACC). If successful, this can expand the understanding of cardiology research among the public and those with low health literacy.
Research Question: Can ChatGPT transform abstracts to be understandable by patient populations who cannot understand medical text without including extraneous or false information?
Goals: We quantified the readability of abstracts published in JACC with clinically verified metrics. We assessed the readability of these same abstracts post-ChatGPT transformation.
Methods: Review article abstracts published in the JACC from January to June 2023 were simplified using ChatGPT. The following prompt was used in ChatGPT to transform abstracts: “Please rewrite the following text to be readable at a 5th-grade level. Do not include information not contained in the original text, and do not exclude information contained within the original text.” Readability was quantified using Flesch-Kincaid Grade Level (FKGL) and Reading Ease (FKRE) scores. A board-certified cardiologist assessed content validity and ensured accuracy.
Results: All 24 abstracts had significantly lower FKGL and higher FKRE scores (P<0.001) after transformation by ChatGPT, indicating improved readability. No extraneous information was found in 24/24 transformed abstracts.
Conclusion: Medical literature exceeds the average reading and comprehension level of US adults, especially among those with low socioeconomic status. Our results indicate that ChatGPT effectively simplifies complex cardiology literature, maintaining information accuracy. This opens the door for further exploration of ChatGPT's role in improving accessibility to various cardiology topics.