What is it about?
This study compared three ChatGPT-based AI models for their ability to support diabetes self-management: Diabetes Self-Management GPTs Support System (a customized model built with diabetes guidelines) GPT-4 Omni (a state-of-the-art, general-purpose AI) GPT-o1 Preview (a safety-focused model) The researchers tested the AI in four scenarios: Insulin administration An older patient with visual impairment A child facing stigma at school A “sick day” scenario requiring special adjustments Each AI’s responses were evaluated for detail, accuracy, emotional tone, and patient-centeredness.
Featured Image
Why is it important?
Diabetes is challenging to manage, especially for older adults, children, and those with complications. AI could reduce burden by giving patients practical, empathetic, and accurate guidance. The three AI models showed clear differences: GPT-4 Omni was the most detailed but sometimes overwhelming. Diabetes Self-Management GPTs was the most empathetic and concise but less deep in complex cases. GPT-o1 Preview was the safest but offered limited guidance. Implication: AI tools must be selected and customized based on patient needs to maximize effectiveness.
Perspectives
Stratified implementation: Different AI models may be better suited for different groups—concise and empathetic systems for the general public, more detailed ones for high-risk or complex cases. Empathy matters: Especially for children and older patients, supportive language is as important as medical accuracy. Public health potential: AI systems could help address stigma, provide education, and extend support in resource-limited settings. Safety and ethics: Human oversight, regulatory frameworks, and privacy protections are essential for responsible use. Future directions: Real-world testing, age-appropriate communication design, and continued refinement of customized AI will be key.
Prof Takuya Omura
National Center for Geriatrics and Gerontology
Read the Original
This page is a summary of: Comparative Analysis of ChatGPT-Based Artificial Intelligence for Diabetes Self-Management Support: Potential of Artificial Intelligence Stratified Implementation and Customization, Diabetes Technology and Obesity Medicine, March 2025, Mary Ann Liebert Inc,
DOI: 10.1089/dtom.2025.0004.
You can read the full text:
Contributors
The following have contributed to this page
