Modelling of glucose-insulin dynamics from low sampled data ?

摘要

In this paper we focus on modelling the glucose-insulin dynamics in the human body for the purpose of controlling the glucose level. Due to the fast dynamics in the glucose-insulin system compared to the natural sampling period (24 h) in a clinical situation, the model structure has to be adapted adequately. This results in a reduced order model with a nonlinear output relation. The development of the estimation methodology is based on a simulation study with a continuous time model. The resulting model structure is used for estimating the parameters of the non-linear system, representing the slow dynamics observed from the slow and sparse sampled clinical data.