Blood Glucose Regulation in Type 1 Diabetes Through Optimized Nonlinear Control Strategies.

The loss of insulin-producing $\beta$ cells in the pancreas characterizes type 1 diabetes, necessitating exogenous insulin therapy for those affected. The objective is to develop a completely automated artificial pancreas. A control system capable of stabilizing the system requires closed-loop strategies. The modified Hovorka model is a detailed nonlinear model able to describe blood glucose regulation through insulin injection. In the proposed work, this well-established model has been adopted to design a double integral fast terminal sliding mode controller using a genetic algorithm for parameter optimization. Different fitness functions have been investigated to apply a genetic optimization algorithm. A comparison with baseline control methods from the literature has been also performed showing a significant improvement.