Artificial Pancreas: from in-silico to in-vivo

Abstract Type 1 diabetes is a disease caused by an autoimmune reaction. The Artificial Pancreas (AP) is an automatic closed-loop system composed of a subcutaneous glucose sensor, a subcutaneous insulin pump and a device on which a control algorithm and a human interface are implemented. The last years have seen an accelerated improvement of these three components that became more reliable and compact, making the system safer, wearable, and usable in real life. An overview on AP and its components is presented together with an introduction on the in-silico tools used to develop and tune the control algorithm and to make pre-clinical tests. Particular attention is devoted to the design of a Model Predictive Control, to the choice of the model and of the constraints, and to the definition of the most relevant performance indices. Most of the choices have been driven by the experience gained by both in-silico and in-vivo trials. In-silico experiments involved thousand of hours of simulations on the Food and Drug Administration accepted simulator equipped with 100 adult virtual patients. In-vivo experiments, of which a complete list is presented, involved about forty thousand hours of trials, first, conducted in a clinical environment and, then, at home.