Glycemic Control Strategies Based On Individual Clinical Parameters

For the patients with type1diabetes mellitus, maintaining good glycemic control has its significance. Technological advancements have paved the way to the closed-loop artificial pancreas, while the continuous glucose monitor and continuous subcutaneous insulin infusion pumps currently help type1patients in controlling glucose levels. However, the full participation is still required in current treatment to prevent hypoglycemia as well as hyperglycemia, which assuredly adds the tremendous burden on patients. Closed-loop insulin delivery systems include a continuous glucose monitor system, continuous subcutaneous insulin infusion pump, and an automated control algorithm. The main challenge of artificial pancreas study is the design of the closed-loop control algorithms, which face enormous difficulties.The traditional open-loop treatment is relatively mature, where the widely used clinical parameters have extremely important practical significance. This paper discusses the significance of these clinical parameters, analyzes the challenges faced in the control algorithm design and proposes two types of fully closed-loop control algorithms for artificial pancreas:one is based on PID control method, and the another is based on model predictive control (MPC) method.The former control scheme implements the control action just based on the input and output data, considering the plentifully rapidity and sufficiently safety. In addition, some clinical rules are given for the controller parameter adjustment and the feasibility of closed-loop application has been investigated. The latter control scheme tries to obtain an individualized MPC for artificial pancreas by employing a so-called expected insulin calibration mode (EICM), which is a helpful tool in the transition from conventional open-loop insulin pump therapy to closed-loop insulin delivery system, as well as in implementing day-by-day tuning of the MPC controller.The simulation results demonstrate the superior performance of these two control systems in hypoglycemia prevention and postprandial hyperglycemia reduction. In addition, the simplicity of the algorithms will benefit physicians with less professional control knowledge.