The top journal of the achievement journal, how to accurately regulate blood sugar in patients with type 2 diabetes? Zhongshan Hospital breaks through through the use of AI systems
Recently, a team led by Li Xiaoying and Chen Ying from the Endocrinology Department of Zhongshan Hospital affiliated with Fudan University, together with a team led by Professor Wang Guangyu from Beijing University of Posts and Telecommunications, published their research results online in the top international medical journal Nature Medicine. For the first time in the world, the joint team proposed to adopt AI system "RL-DITR" based on reinforcement learning algorithm to formulate insulin decision-making strategies, effectively improving the accuracy of insulin treatment programs for patients with type 2 diabetes. The research results can provide personalized and dynamic diagnosis and treatment programs for patients with type 2 diabetes, assist in establishing a hierarchical diagnosis and treatment system, and improve the efficiency of chronic disease management.
China is the largest country of diabetes. According to the latest data, one in every nine adults has diabetes, of which type 2 diabetes accounts for more than 90% of the total number of diabetes, and nearly 50% of patients need insulin injection treatment.
How to accurately and efficiently adjust insulin dosage for a large group of diabetes patients? This has always been a difficult problem that has plagued the medical community.
How to accurately and efficiently adjust insulin dosage for a large group of diabetes patients? This has always been a difficult problem that has plagued the medical community.
![The top journal of the achievement journal, how to accurately regulate blood sugar in patients with type 2 diabetes? Zhongshan Hospital breaks through through the use of AI systems](https://a5qu.com/upload/images/bd98b30c16536aac072b4fa2fe8beb03.jpeg)
This system can predict the optimal drug dosage in real-time based on the patient's historical data and current physiological conditions, as well as the differences in insulin response among different patients and changes in insulin demand during disease progression. It can develop personalized, accurate, and dynamic treatment strategies to achieve blood glucose control goals.
Traditional insulin dose adjustment mainly relies on the experience of doctors and cannot meet the dynamic changes between individuals. Since 2020, Li Xiaoying, Chen Ying and Professor Wang Guangyu have jointly carried out the research on AI system "RL-DITR" based on innovative algorithms such as reinforcement learning to optimize insulin treatment for patients with type 2 diabetes.
This system can predict the optimal drug dosage in real-time based on the patient's historical data and current physiological conditions, as well as the differences in insulin response among different patients and changes in insulin demand during disease progression. It can develop personalized, accurate, and dynamic treatment strategies to achieve blood glucose control goals.
![The top journal of the achievement journal, how to accurately regulate blood sugar in patients with type 2 diabetes? Zhongshan Hospital breaks through through the use of AI systems](https://a5qu.com/upload/images/2aadfeb2437d06cc5f2cf47b672d4ecc.jpeg)
Gu Jianying, Secretary of the Party Committee of Zhongshan Hospital Affiliated to Fudan University, stated that building a digital healthcare world and a new ecosystem of medical services is the hospital's pursuit. The hospital deeply integrates new technologies such as 5G, artificial intelligence, big data, and digital twins with high-quality medical resources to create the first "5G+digital twin smart medical ecosystem" in the country.
Research has found that compared to other artificial intelligence models and current clinical standard protocols, RL-DITR is closer to the judgment of doctors with rich clinical experience. Compared with their recommended insulin dosage, the difference is only 1.2 units. At the same time, it increases the percentage of time for patients to meet glucose standards by 24.1% and does not cause serious consequences such as hypoglycemia or ketoacidosis.