Recently, an international team composed of researchers from China and Singapore successfully developed the world's first multi-modal artificial intelligence system for diabetes diagnosis and treatment, DeepDR-LLM. This groundbreaking research was published in Nature Medicine on July 19, 2024.
This pioneering study was led by Professor Wei-Ping Jia and Professor Hua-Ting Li from the Sixth People's Hospital affiliated with Shanghai Jiao Tong University School of Medicine. They closely collaborated with Professor Bin Sheng's team from the Department of Computer Science, School of Electronic Information and Electrical Engineering at Shanghai Jiao Tong University, Professor Tian-Yi Huang's team from Tsinghua University, and Professor Yu-Zong Qin's team from the National University of Singapore to complete the research.
Image source note: The image was generated by AI, and the image licensing service provider is Midjourney.
The DeepDR-LLM system integrates large language models and deep learning technologies to achieve multi-modal generation functions for medical imaging diagnosis and treatment advice. This system not only provides auxiliary diagnosis for diabetic retinopathy but also offers personalized comprehensive management suggestions for diabetes.
To verify the effectiveness of the system, the research team conducted multi-center retrospective validation in seven countries across Asia, Africa, and Europe. Additionally, they carried out prospective real-world validation in primary healthcare institutions in China, providing high-quality evidence-based support for the system's application in the field of diabetes medicine.
The research team stated that the DeepDR-LLM system is expected to significantly enhance the efficiency of screening for diabetic retinopathy and the level of diabetes management at the primary care level, providing an innovative digital solution for global diabetes prevention and control.