ABSTRACT
When checking for identifiability caused by the inherent model structure it is recommended that the symbolic method is used, as this method is accurate, can be used to find estimable parameter combinations in parameter redundant models and can be used to generalise results. If the model is structurally too complex to obtain the rank of the appropriate derivative matrix, one option is the extended symbolic method. An alternative is the hybrid symbolic-numerical method, which is much simpler to use. Using the extended symbolic method can also obtain the estimable parameter combinations in parameter redundant models and can be used to generalise results. When checking for identifiability for a specific data set (with a particular model) the hybrid symbolic-numeric method is ideal. It is more accurate than than the Hessian method, or other numerical methods, and can be added to a computer package to provide automatic identifiability results.
