Intelligent Control of an Artificial Pancreas using Artificial Neural Networks
Artificial pancreas. Nonlinear control, Feedback linearization, Intelligent control, Artificial neural networks, Radial basis functions
Type 1 Diabetes Mellitus is a disease that affects millions of people around the world. Recently, the incredible progress of embedded devices has given rise to proposals of devices that pump insulin subcutaneously, with the purpose of automatically regulating blood glucose level in diabetic patients. This way, the Artificial Pancreas could provide a better quality of life with more autonomy and comfort to the patients. The goal of this work is to design a nonlinear controller with feedback linearization and a radial basis function artificial neural network as an uncertainty estimator for an artificial pancreas. The IVP model for blood glucose regulation is used to simulate the dynamics of the virtual patient. A 7-day, 3-meal per day simulation is performed on 20 virtual patients. The virtual patients are generated in a random manner, following a normal distribuition, with the goal of introducing interpatient variability to the simulation. Also, the parameters for each patient vary over the course of the simulation in a sinusoidal way, with the goal of including intrapatient variability in the simulation. The proposed controller neither has knowledge of the system dynamics nor is alerted when a meal is eaten. After simulating 20 patients, the controller reached a mean time in normoglycemic regimen of over 96%, indicating its ability to estimate uncertainties of not only external disturbances but also unmodeled dynamics. Furthermore, none of the patients reached severe hipoglycemic or hiperglycemic levels, minimizing hipoglycemic episodes on postpandrial periods.