S1114 Artificial Intelligence in GI Education: ChatGPT’s Role in Gastroparesis Awareness

Introduction: Gastroparesis is a chronic gastrointestinal motility disorder characterized by delayed gastric emptying in the absence of mechanical obstruction. As patients increasingly rely on generative artificial intelligence (AI) tools for health information, it is essential to evaluate the accuracy of such platforms. ChatGPT, a publicly accessible text-generating AI (TGAI), is among the most widely used models for patient-centered inquiries. This study assessed the reliability and consistency of ChatGPT’s responses to commonly asked questions about gastroparesis. Methods: Fifty frequently asked patient questions were developed by internal medicine physicians based on clinical experience and literature review. Each question was submitted to ChatGPT (GPT-4, OpenAI, April 2025 version) in 3 independent sessions. Ten evaluators—7 internal medicine residents and 3 faculty physicians—reviewed the responses and rated them as Appropriate (accurate and complete), Inappropriate (partially accurate or incomplete), or Unreliable (factually incorrect or misleading). Discrepancies were resolved through group consensus. Results: Among a total of 150 responses (50 questions submitted in 3 separate ChatGPT sessions), 138 (92%) were rated as Appropriate, 9 (6%) as Inappropriate, and 3 (2%) as Unreliable. ChatGPT demonstrated high accuracy in addressing topics related to disease pathophysiology, diagnostic strategies, and pharmacologic management. The most frequent limitations were seen in responses involving dietary guidance and psychosocial considerations, which often lacked individualization or depth. Conclusion: ChatGPT delivered a high proportion of clinically accurate responses to common patient questions about gastroparesis, highlighting its potential as a supplementary tool for patient education. However, occasional inaccuracies—particularly in nuanced or individualized topics—underscore the necessity of physician oversight. These findings support the cautious integration of AI technologies into clinical communication, with emphasis on clinician validation to ensure information is guideline-concordant, safe, and tailored to each patient's needs.