Q: What should I do when the fit goes on forever?

A: The reason for this is that the optimization algorithm used in the program has trouble converging, given the model and the data. Usually, this happens because the data are not informative enough and/or the model is too complex to guarantee identifiability.

Possible solutions are:

1. Check the a priori identifiability of your model: it might be that your model is not a priori identifiable from data, that is, the solution for the unknown parameters of the model is not unique, given the model structure and the experimental design. For a formal definition and more information about a priori identifiability and software tools to check for it.
   E.g, see S. Audoly, L. D'Angio', M. P. Saccomani and C.Cobelli, "Global Identifiability of Linear Compartmental Models - A Computer Algebra Algorithm", IEEE Transactions on Biomedical Engineering,45: 1, January 1998.
2. If you have some additional information about the parameters, make them Bayesian parameters by giving SAAM II their Population Mean and Standard Deviation.
3. Simplify the model structure, deleting nonaccessible compartments and creating links between the parameters when appropriate.
4. Redesign the kinetic experiment and collect additional data, then attempt again to fit your model.

See also the links:

1. What should I do when the covariance matrix is unreliable?
2. When and how should I use Bayesian fitting?

for more information on how to deal with this situation.

<< Back to Support Page