Insulin Dosage Optimization Based on Model Predictive Control with the Model Parameters Estimation

Abstract

Diabetes is a disease characterized by the inability of the pancreas to regulate blood glucose concentration. The development of the closed-loop control method for the treatment of type 1 diabetes is a highly desired endeavor for patients, physicians and scientists. Focusing on the need for maintaining normal glycaemia, the paper proposes an integrated closed-loop control method, which is the model predictive control (MPC) method combined with model parameters estimation. Using the static glucose model to be the internal model can’t achieve the good performance in different objects because of the individual differences. By updating the internal controller model with current measurement information using the parameters estimation, mismatch between the actual controlled object and the internal controller model is significantly reduced. Therefore, the predictions of future glucose values using the updated model are more accurate than those of the static input–output model. Finally, the integrated control method was tested on the T1DM simulator which was accepted by the FDA. Four objects are selected randomly in T1DM simulator. The control results showed it can modulate the glucose concentration to 85~120 mg/dl without severe hypoglycemia in the effective time. In conclusion, it is promising for the control of glucose concentration in subjects with type 1 diabetes.