Self-tuning controller for regulation of glucose levels in patients with type 1 diabetes

Closing the loop with a fully automated artificial pancreas will definitely improve the life of patients with type 1 diabetes. An adaptive control strategy is proposed in order to dynamically respond to unpredicted glucose fluctuations due to internal or external perturbations. The adaptability of the controller is further assured with models derived from continuous glucose measuring device data collected from the patient. The implicit self-tuning tracker is used to keep the glucose concentrations within normoglycemic range, while the controller parameters are directly tuned at each step based on patient specific time-series models. Closed-loop results are demonstrated on a simulated patient assuming intravenous measurement of glucose and subcutaneous infusion of rapid-acting insulin.