This research introduces a framework for training and validating diabetes-specific large language models that enhance multiple aspects of diabetes care.

• Developed a comprehensive data processing pipeline for diabetes-related content
• Created specialized LLMs that address diverse diabetes tasks across stakeholders
• Validated effectiveness across medical, educational, and support applications
• Demonstrates potential for personalized healthcare through task-specific AI assistance

This work represents a significant advancement in medical AI by focusing on a high-burden chronic disease affecting millions worldwide, creating specialized tools that can improve patient outcomes through better coordination among healthcare providers, patients, and caregivers.

Diabetica: Adapting Large Language Model to Enhance Multiple Medical Tasks in Diabetes Care and Management