Towards Event-Trigger Impulsive MPC for the Treatment of T1DM Handling Limited Resources

Tight control of blood glucose is a difficult task in patients with type I diabetes. Nowadays, the artificial pancreas is becoming an auspicious treatment for them. It consists of an insulin pump, a continuous glucose monitoring (CGM) sensor, and a control strategy running on a portable device. However, it faces many challenging issues, including limited resources such as insulin supply and battery power capacity. This paper examines this problem within the model predictive control (MPC) framework. The standard impulsive zone MPC formulation has been modified by adding an event-trigger feature and two new constraints that help extend the battery life and the insulin reservoir. The event-trigger scheme allows the new formulation to decide at which time instants it is necessary to compute a control action. In addition, when battery life and insulin supply are below the established limits, the two constraints become active, providing more time to the treatment than the standard strategy. The performance of this new proposal is assessed through several simulations, obtaining promising results that deserve further research.