Adjustment of Sensitive Parameters of a Mathematical Model of Glucose Metabolism Using an Evolutionary Algorithm

Abstract

Currently, closed-loop therapy is a promising option for individuals with type 1 diabetes mellitus. It requires a glucose metabolism model and a control algorithm to reach a suitable automatic insulin delivery. Many mathematical models have been proposed considering the glucose-insulin dynamics and nominal parameters of a standard patient. Nevertheless, the open-loop therapy based on continuous glucose monitoring (CGM) and continuous subcutaneous insulin infusion (CSII) provides sufficient data to seek patient-oriented models instead of the nominal ones. In consequence, patient-oriented models could open the possibility of having personalized closed-loop therapies. In this work, a methodology for automatic adjustment of sensitive parameters of a physiological mathematical model of glucose metabolism is presented. The methodology is based on an evolutionary algorithm and it is exemplified with a physiological model. This methodology provides a criterion to select the best model which can reproduce the glucose-insulin dynamics of a diabetic patient according to the data provided by the CGM and CSII systems.