Introduction and Objective: Type 2 diabetes (T2D) care involves multifactorial clinical decisions that integrate comorbidities, weight, safety, glycemia, adherence, and cost. Guideline updates and expanding trial evidence increase burden and variation in treatment decisions. We developed a diabetologist workflow-aligned agentic AI providing evidence-cited recommendations. Methods: The system generates an evidence-cited report aligned with diabetologist workflow, detailing clinical reasoning for patient triage/problem list, medication recommendation, treatment strategy, dose adjustment, and monitoring/education. Synthetic T2D cases were developed and validated for clinical relevance by 3 senior diabetologists. 12 diabetologists performed double-blind qualitative assessment on AI recommendations for synthetic cases (N=48) using a Delphi-verified 29-item assessment framework. Results: On a 5-point Likert scale, the proportion of ratings ≥4 were: patient triage/problem list 96.1%, medication recommendation 90.9%, treatment strategy 85.4%, dose adjustment 89.2%, monitoring/education 87.8%, reasoning reliability 100%, clinical utility 97.0%, real-world feasibility 90.9%. Conclusion: The agentic AI produced evidence-cited recommendations with high clinician acceptability across core T2D decision components. The system may help standardize treatment decisions in real-world clinical practice. Disclosure S. Baek: None. J. Kim: None. S. Jin: None. G. Kim: None. Y. Lee: None. J. Kim: None. S. Cho: None. R. Oh: None. B. Kim: None. M. Jang: None. S. Ko: None. M. Moon: None. K. Kim: None. K. Hur: None. Funding Future Medicine 2030 Project of the Samsung Medical Center (#SMX1250111); The Institute of Information & Communications Technology Planning & Evaluation (IITP) grant funded by the Korea government (MSIT) (No. RS-2024-00357879).
BAEK et al. (Fri,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: