Clone wiki

D2D Software / Uncertainty analysis

Uncertainty analysis and experimental design

The software implements the profile likelihood approach

This general approach allows to infer both the structural and the practical identifiability of parameters in non-linear and possibly dynamic models by calculating the profile likelihood. Furthermore, it can be used to calculate likelihood-based confidence intervals and to design optimal experiments that improve parameter identification and therefore also the predictability of a model.

Alt text

An application to a model from cell biology (Becker et al., Science 2010, see in example applications) that illustrates the iterative cycle between modeling and experimentation can be found in:

A more general overview about identifiability and its consequences on model predictions in terms of observability can be found in:

The results obtained by the profile likelihood approach were compared to results of Markov-chain Monte Carlo sampling in:

The profile likelihood approach was extended to cover arbitrary model predictions in:

A general overview about the profile likelihood methodology is given in:

See also Profile Likelihood Approach for questions concerning the profile likelihood approach.

Updated