Selection Range of the Control Weighting Parameter for a Closed-loop Artificial Pancreas Based on Generalized Predictive Control

Abstract

The artificial pancreas (AP) system is a promising approach to maintain the blood glucose concentration (BGC) of type 1 diabetes (T1D) patients in a euglycemic range (70-180 mg/dl). We have developed an AP based on generalized predictive control (GPC) and found that its parameter settings should be different for adult and adolescent patient groups, like softening factor. Here, the selection range of another important factor, control weighting parameter (λ), was studied for adult and adolescent group, respectively. Tests with the UVA/Padova type 1 diabetes mellitus simulator (T1DMS) approved by the US Food and Drug Administration (FDA) showed that the valid and safe range of λ for the GPC controller with the adult patient group was 2 to 8. But it should be cautious to select a λ value for a GPC-based AP with adolescent patients because their hypoglycemia risks rapidly increased with the increasement of λ. There was a much smaller selection range of λ for the GPC controller with the adolescent patient group and 2 was proposed as a valid and safe value here. Those results are helpful to the AP design and diabetes treatment, especially for adolescents with T1D.