Bio-Inspired Strategies for Modeling and Detection in Diabetes Mellitus Treatment

5: Neural model for glucose–insulin dynamics

Abstract

Obtaining models using deep neural networks plays an important role in modern science and technology due to the need for synthesis of models that describe the dynamics of unknown systems in a precise and reliable manner. Obtaining these neural models is carried out through data obtained from input–response experimentation of the system. Although the data may contain noise, neural networks are capable of modeling the dynamics with high precision. This work shows identification of models of glucose–insulin dynamics offline using deep neural networks, in addition to online identification using the discrete high-order neural network. Neural networks prove to be a powerful tool for identification. ...