The expert team formed by the MIFA Laboratory of Shanghai Jiao Tong University's Qingyuan Institute and the Endocrinology Department of Zhongshan Hospital affiliated with Fudan University has recently co-developed a diabetes-specific large model named Diabetica. This model integrates the language processing capabilities of large models with specialized diabetes knowledge, aiming to provide intelligent support for doctors, patients, and medical education.
The team adopted a new framework to establish a comprehensive data processing system, creating high-quality datasets and evaluation benchmarks. They fine-tuned the large language model, significantly enhancing its capabilities in handling diabetes-related tasks. Clinical studies show that Diabetica has potential applications in diabetes care, including providing personalized medicine, assisting in medical education, and simplifying clinical tasks.
The team emphasizes that the research introduces the development and evaluation framework of the model, highlighting its significant potential in enhancing clinical practice and providing personalized, data-driven diabetes support.
Paper URL: https://arxiv.org/abs/2409.13191
GitHub URL: https://github.com/waltonfuture/Diabetica