Parameter estimation of glucose-insulin dynamical model using an extended kalman filter

Diabetes Type 1 occurs when the body fails to produce insulin due to an attack by the immune system on the β-cells. This disease caused an increase in blood glucose that can lead to potential health complications. The Bergman’s Minimal Model (BMM) and Extended Bergman Minimal Model (EBMM) is used to understand the interaction between glucose and insulin. Meanwhile, some parameters cannot be measured directly, so Extended Kalman Filter (EKF) is used to estimate the parameters. EKF, the development of Kalman Filter algorithm, is a nonlinear estimator that consists of two stages that is predict and update stage. In this research, five dataset blood glucose level of diabetic patients type I is used to estimate the value of insulin-independent glucose-utilization rate and glucose effectivity level. The simulation shows that estimation data is close to actual data with small errors 2.36 × 10-4. It shows that EKF method has produced effective results in providing estimated parameters of BMM and EBMM models.