Ensemble Model Predictive Control Strategies Can Reduce Exercise Hypoglycemia in Type 1 Diabetes: In Silico Studies

This contribution presents an individualized Ensemble Model Predictive Control (EnMPC) algorithm for blood glucose (BG) stabilization and hypoglycemia prevention in people with type 1 diabetes (T1D) with detectable patterns of exercise. The EnMPC formulation can be regarded as a simplified multi-stage MPC that considers $N_{en}$ future exercise patterns identified from the patient's recent behavior. The control action is determined from a consensus across the ensemble, where each scenario is treated by a specific MPC algorithm, accounting for the underlying disturbance likelihood. The patient's physical activity behavior is characterized by an exercise-specific input signal derived from the convolution of patients exercise records, e.g., obtained from activity monitors, and physiological impact curves from the literature. The proposed EnMPC strategy was tested on the complete in silico adult cohort of the FDA-accepted UVA/Padova metabolic simulator. Results confirm a tangible improvement in time spent below < 70 mg/dL ($\mathbf{p} < 0.001$) without increasing hyperglycemia after 30 min of moderate exercise in comparison to a similarly tuned MPC baseline controller (rMPC). Additionally, there was a significant reduction in the number of hypotreatments ($\mathbf{p} < 0.001$) that the patients received during and after exercise between the two controllers.