Possible ways to test SimBiology model robustness

I already tried adding noise to the experimental data; to test the model robustnessl. However, i am just wondering if there are other feasible way to check if a SimBiology built model is robust or not.

I will apprecite any idea on that. Thank you

Blessing.

[Arthur Goldsipe]

Can you say more about what you mean by robust? I'm not sure if it's relevant to you, but the SimBiology Model Analyzer supports both local and global sensitivity analysis, which can help you understand how a model might behave differently as you adjust the parameter values.

[BLESSING AKAH]

Model robustness implies that, the built model should be able to fit any experimental data, with different covariants. The model is independent on the experimental data. For example: tumor growth modeling for combination therapy, the model should be able to fit any experimental data derived from pateints with different covariants, drug concentration, etc.

When the model prediction did not match the experimental data, then such model is not robust; it is built specifically for a particular data set.

I know of sentivity analysis, but that does not show if a model is robust or not. It only indicates how each parameters affects the tumor growth. From there, one can know the parameters with greater or least effect.

I am looking at something more sophisticated.

Answer 1

[Arthur Goldsipe]

The closest thing I can think of off the top of my head is cross-validation. Maybe you want to take a look at this page and see if anything looks useful.