Robust glucose control via μ-synthesis in type 1 diabetes mellitus

Abstract

Diabetes Mellitus (DM) is a disease which afflicts a large group of the world population. It affects people from any socio-economic status, physical characteristics, age or gender. This pathology is a major public health problem and it has been focused as a disease requiring priority attention. In this way, the idea of an artificial pancreas has become one of the areas of research for multiple disciplines because it offers a promising approach for Type 1 Diabetes Mellitus (T1DM) treatment. The aim of this artificial pancreas is to reduce disease complications despite the variability of internal and external body conditions. This variability can be treated as parametric perturbations or uncertainties. In this contribution a sensitivity analysis is performed on the mathematical model that represents a T1DM subject, resulting in six parameters as the most representative to characterize parametric uncertainty. The work is focused on the synthesis of a robust controller using the μ-synthesis technique in order to obtain a glycemic controller for a subject with T1DM. Moreover, a new transfer function Ps was identified in order to describe the characteristics of the TheraSense FreeStyle Navigator glucose sensor. This last was employed in controller design. The performance of this controller was tested in the well-known UVa/Padova simulator.