An Approach for Artificial Pancreas to Control the Type-I Diabetes Mellitus

Abstract

Currently diabetes mellitus is worldwide issue and challenges for diabetes community for healthy life. An effort is made to develop the idea of getting a fully automated artificial pancreas. The artificial pancreas is a developing technology to help patients with diabetes of type 1 diabetes mellitus (T1DM) to control automatically their blood glucose level by making available the alternative endocrine functionality of a healthy pancreas. The concept of controllability and observability for the linearized control system of human glucose insulin system is used so that we can have a feedback control for artificial pancreas. For the purpose of automatic artificial pancreas in the glucose regulatory system, we consider the Glucose Insulin (GI) Systems and Glucose Insulin Glucagon (GIG) systems. These models includes Augmented Meal Model (AMM), Reduced Meal Model (RMM), fractional order glucose insulin system, a composite model of Glucagon-Glucose Dynamics Model and Sorenson model being comprehensive model for Type-1 Diabetes Mellitus (T1DM). These models can be used to simulate a glucose insulin system for the treatment of T1DM. The Lyapunov Equation is used to check the stability analysis of the model. A fractional-order time derivatives model is presented for comprehensive glucose insulin regulatory model. A fractional-order state observer is designed for approximating the structure of a blood glucose-insulin with glucose rate disorder to show the complete dynamics of the glucose-insulin system with the fractional-order at α ∈ (0 < a < 1]. The developed method provides the observer estimation algorithm for a glucose-insulin system with unknown time-varying glucose rate disturbance. Numerical simulations are carried out to demonstrate our proposed results and show the nonlinear fractional-order glucose-insulin systems are at least stable as their integer-order counterpart in the existence of exogenous glucose infusion or meal disturbance. Controllability and observability of the linearized model are calculated under two different cases,for case 1 insulin is taken as an input and case 2, insulin and glucagon are taken as an input for the system. This played an important role in the development of fully automatic artificial pancreas by stabilizing the control loop system for the glucose-insulin glucagon pump. Proportional Integral Derivative (PID) controller is designed for an artificial pancreas by using the transfer function. According to the desire value, the algorithm of an artificial pancreas measures the glucose level in the blood of a patient by using glucose sensor that sends a signal to an insulin glucagon pump to adjust the basal insulin. A closed-loop system is tested in simulink environment and simulation results show the performance of the designed controller. We convert the Sorenson model to Sorenson model type 1 diabetes mellitus because this is the most comprehensive model in the Glucose Insulin Glucagon dynamics for human. This may play an important role in the development of fully automatic artificial pancreas and stabilize the control loop system for the Glucose Insulin Glucagon pump. It would be helpful for type 1 diabetic patients to control their diseases. The thesis is also review the state of art in hypoglycemia prevention and detection technique in the closed-loop artificial pancreas. Hypoglycemia is the major adverse effect of insulin therapy and therefore minimizing the risk of hypoglycemia, by applying different control and detection techniques is often considered in the development of artificial pancreas.