An Adaptive Predictive Control with Robust Filter Approach to Blood Glucose Regulation in Diabetics

G. CAMPETELLI; D. ZUMOFFEN; M. BASUALDO

Oro Verde, Entre Ríos

Congreso; XIV Reunión de Trabajo Procesamiento de la Información y Control -- RPIC 2011; 2011

Universidad Nacional de Entre Ríos

An artificial pancreas to solve the problem of type I diabetes mellitus is a closed loop system composed of a blood glucose biosensor, a control algorithm and an insulin pump. It is in charge of maintaining plasma glucose concentration of a diabetic patient within the healthy range of 70 - 180 mg/dl. Even so, the process of blood glucose homeostasis is time-dependant, mainly because of the daily variations of insulin sensitivity amongst many other factors. Accounting this important variability it was thought that an Adaptive Predictive Control with Robust Filter (APCWRF)couldrepresent a good alternative for regulating blood glucose. Hence, this is the main contribution of this paper since there is no report about APCWRF applications on this complex problem. It is tested on a recognized simulator of type I diabetic patients. The controller uses a Finite Impulse Response (FIR) model for the predictions. The performance of the APCWRF is compared to that of a Generalized Predictive Controller (GPC) using the same nominal internal model without parameters adaptation. Both controllers are improved with a feed-forward compensation for meal intakes. From the comparative results can be concluded about the real advantages of the APCWRF for dealing with the regulation problem of an average adult type I diabetic patient.