ChatGPT's Role in Coronary Artery Bypass Graft Information: A Critical Assessment

This research evaluates the accuracy and reliability of ChatGPT's responses to inquiries concerning coronary artery bypass graft (CABG) surgery.

Between December 1, 2024, and December 15, 2024, two sets of questions were used to assess ChatGPT's performance: (1) a list of frequently asked questions (FAQs) sourced from official cardiovascular surgery websites, professional healthcare platforms, and social media, and (2) a list of scientific questions derived from the 2021 American Association for Thoracic Surgery (AATS) Guidelines for CABG in patients with ischemic cardiomyopathy and heart failure. Responses were evaluated using the modified DISCERN scoring system. To test reproducibility, each question was asked twice on separate computers within the same day. Two experienced cardiovascular surgeons independently scored the responses, and any disagreements were resolved through discussion.

A total of 107 FAQs were assessed, of which 88 met the inclusion criteria. Based on the modified DISCERN scale, 71 responses (80.7%) received a score of 5, while 9 responses (10.2%) were rated as 4. The highest accuracy was observed in the preoperative preparation category, where all responses achieved a perfect score. Regarding scientific questions derived from the AATS Guidelines, 15 out of 20 (75.0%) were answered thoroughly and satisfactorily. The reproducibility rate was 92.0% for FAQs and 90.0% for guideline-based questions.

This study is the first to demonstrate that ChatGPT provides highly accurate and dependable responses regarding CABG surgery, particularly for frequently asked questions. Although performance declines slightly when addressing scientific questions based on guidelines, the reproducibility rate remains high. These findings indicate that AI-driven tools like ChatGPT could play a valuable role in patient education and enhancing public awareness about CABG surgery.