Efficacy and Safety of an Artificial-pancreas-like Learning-based Control in Type 1 Diabetes on Multiple Daily Injection Therapy

The AP-A Dosage Decision Support System represents a new solution in diabetes care, offering customized insulin dosage recommendations to patients. This system is a product of integrating four pivotal modules: the individualized model learning module, the risk-sensitive control module, the Bayesian optimization module, and the safety constraint module. Together, they establish a robust framework that employs advanced computational methodologies to deliver precise and personalized insulin dosage guidance, significantly improving the effectiveness and safety of diabetes treatment plans. The actual injection dose in the intervention group was executed by the doctor after approval based on the recommendation of the AP-A Dosage Decision Support System.