(P619) Generative AI in T1D: A Cross-Platform Analysis of Recommendation Consistency and Clinical Concordance

The rapid advancement of generative AI tools and their potential for widespread adoption in healthcare settings have created an urgent need to understand their reliability and consistency in diabetes care recommendations before their widespread clinical implementation. As people with diabetes increasingly turn to generative AI platforms for health information and diabetes specialists consider these tools as decision-support aids, the gap in evidence regarding their accuracy and consensus presents a significant safety and quality concern. This poster shares findings from a systematic evaluation of the ability of four major generative AI platforms to provide consistent, evidence-based T1D management recommendations. Attendees will learn about our methodology for testing internal consistency within platforms and external consensus across platforms using real-world clinical cases. The authors detail their approach to comparing AI-generated recommendations against established clinical standards and experienced clinician judgment across seven critical domains of diabetes care, providing DCESs with some initial evidence and thoughts to inform decisions about AI integration in their practice. Authors: Geanna Miranda, OMS III; Clipper F. Young, PharmD, MPH, CDCES, BC-ADM, BCGP, APh, FADCES, Touro University California College of Osteopathic Medicine, Department of Clinical Sciences and Community Health. Funding Source: Touro University California College of Osteopathic Medicine