Exploring Robustness in Blood Glucose Control with Unannounced Meal Intake for Type-1 Diabetes Patient

The importance of efficient diabetes treatment and its' reliable automation is rising with the prevalence of this chronic condition worldwide. Robustness is one of the enablers of the safe automation of plasma glucose control, for it can ensure consistent behavior when the controlled dynamics are changing or partially unknown. Hence, this work focuses on assessing the capabilities of a robust nonlinear controller framework in a simulated environment. Linear Parameter -Varying modeling is combined with robust control techniques, supported by an Unscented Kalman filter. These controllers are then subjected to additional constraints in search of a practical trade-off between disturbance rejection and the severity of transient behavior. Simulations with unannounced meal intakes explore and compare the effect of these additional constraints and configurations using the virtual patients provided by a well-known in silico simulator. The simulation results indicate that this control method can ensure adequate blood glucose control and has the potential to support other control algorithms to realize a safe and reliable artificial pancreas.