Insulin evaluation based on reinforcement learning

摘要

The optimal insulin delivered by the insulin infusion pump is determined by using a reinforcement learning algorithm aimed at personalized glucose regulation. The algorithm optimizes the daily basal insulin rate and insulin to carbohydrate ratio for each patient based on the patient's measured glucose profile. The proposed algorithm can learn real-time patient-specific features captured in daily glucose profiles and provide individual insulin treatments. Automatic and personalized adjustment methods contribute to the optimization of performance algorithms.