Resumen de Contributions to modelling and control for improved hypoglycaemia and variability mitigation by dual-hormone artificial pancreas systems
People with Type 1 Diabetes lack the ability to secrete insulin and therefore need to regulate their blood glucose with exogenous insulin delivery. The Artificial Pancreas is presented as the ideal technological solution to reach the therapeutic goals of normoglycaemia, freeing the patient from the current burden of self-control and management. Nevertheless, the risk of hypoglycaemia and the high glycaemic variability are still a limiting factors in the current control algorithms integrated in the Artificial Pancreas.
The purpose of the present thesis is to delve into knowledge of hypoglycaemia and to advance in the artificial pancreas control algorithms in order to minimise hypoglycaemia incidence and reduce glycaemic variability. After providing an overview of the state of the art in the eld of glucose control and articial pancreas, this thesis addresses issues on modelling and control, with the following contributions: An extension of the Bergman Minimal model accounting for counterregulatory response to hypoglycaemia is presented. This model explains the relationship between the several physiological changes produced during hypoglycaemia, with adrenaline and free fatty acids as main players. As a result, a better understanding of hypoglycaemia is gained, allowing to explain a paradoxical auto-potentiation of hypoglycaemia as modeled through functional approaches in the widespread used UVA-Padova Type 1 Diabetes simulator, which will be used in this thesis for in silico validation of the developed controllers.
An assessment of glucose variability metrics and control quality indices is carried out. The evaluation of the glycaemic variability on the controllers performance is necessary; but there is not a gold standard variability metrics yet. Therefore, an analysis of the variability metrics available in literature is conducted in order to define a recommendable set of indicators.
Due to the limitations of single-hormone artificial pancreas systems in mitigating hypoglycaemia in challenging scenarios such as exercise, this thesis focuses on the developement of new dual-hormone control algorithms, with concomitant infusion of insulin and glucagon. A coordinated dual-hormone controller with parallel control structures is proposed as a feasible control algorithm for hypoglycaemia mitigation and glycaemic variability reduction, demonstrating superior performance as currently used control structures with independent insulin and glucagon control loops. The controllers are designed and evaluated in-silico under challenging scenarios and their performance are assessed mainly with the set of metrics defined previously as the recommendable ones.
