Model-less versus Model-Based Controllers for Diabetes

In this paper, a closed-loop control system was suggested for continuous insulin infusion to improve the traditional discrete methods and make diabetes therapy automatic in practice. Based on continuous injection model, two controllers were designed to tackle a control problem. The controllers designed in this paper are: model-less controller i.e. proportional integral derivative (PID), and model-based controller i.e. linear model predictive controller (LMPC). Even with serious disturbances to glucose, such as exercise, delay or noise in glucose sensor and nutrition mixed meal absorption at meal times, the linear model predictive controller can perform well in simulation experiments compared to the PID controller. Therefore, the LMPC would be preferable and advisable when it comes to critical diabetic cases.