Model Predictive Control of Type 1 Diabetes

The paper presents an overview of the most important elements for the synthesis of a Model Predictive Control (MPC) algorithm for the development of an Artificial Pancreas (AP). Three possible control schemes for meal compensation are described and compared. MPC individualization is discussed through different solutions. Of paramount importance, for a realistic in-silico test, is the use of the population simulator, that has been accepted by the Food and Drug Administration (FDA) as a substitute of the preclinical animals studies. The concepts and techniques described in the paper are the core of the AP algorithm that has been used in the last 4 years during about 4000h of clinical experiments.