Parameter sensitivity analysis is a commonly used technique for determining parameters of a model to be estimated from data. Differential sensitivity analysis is among the most popular techniques for sensitivity analysis, however, it does have the drawback that it is a local approach, i.e., the results from the sensitivity analysis are only ensured to be accurate at the values of the parameters for which the sensitivity analysis was performed. This by itself presents a problem as sensitivity analysis is to be performed to determine which parameters to estimate from data, while at the same time the results from the sensitivity analysis depend upon the values of the parameters. This paper presents three techniques for analyzing how results from sensitivity analysis change with variations in the nominal values of the parameters. The techniques are based upon performing sensitivity analysis of the parameter sensitivities. All techniques are applied to a model describing a signal transduction pathway in liver cells and the results are interpreted.
Dynamics of Continuous, Discrete and Impulsive Systems 14, No. S2, pp 220-226 (2007)