A New Algorithm to Test Global Identifiability of Physiological Nonlinear Models

https://doi.org/10.1016/S1474-6670(17)44581-2Get rights and content

Abstract

Parameters characterizing the intimate functioning of biomedical systems are usually not measurable. Dynamic state models usually nonlinear are formulated and input-output experiments designed to solve this problem as a parameter estimation problem. A fundamental problem for well posedness of parameter estimation is a priori global identifiability which deals With uniqueness of the solution for the unknown model parameters. In this paper, we study global identifiability of nonlinear dynamic models of biomedical systems. A neew algorithm based on differential algebra is presented Which calculates the exhaustive summary by means the characteristic set associated to the system and allows the study of global identifiability for general model structures and multi input-multi output configurations. Three case studies are presented, a model of drug kinetics in the body, a model of glucose metabolism in the brain and a model of insulin control on body tissue glucose utilization

Keywords

modelling
identifiability
identification
parameter estimation
optimal experiment design

Cited by (0)

View Abstract