I used PRCC in one paper (A model of interactions between radiation-induced oxidative stress, protein and DNA damage in Deinococcus radiodurans, available on my page) because reviewers requested it. But I would be grateful for comments about how useful it is in general and what are some alternatives.
For me PRCC is always the first step in assessing the global sensitivity of a model on the uncertainties in its parameters.
The local analysis (done mainly through calculating/approximating the derivatives with respect to the parameters) provides direct information on the effect of small parameter perturbations about their nominal values, but it does not indicate the effect of concurrent, large perturbations in all model parameters (in biomathematics that's almost always the case).
In order to calculate PRCC you simulate exactly those simultaneous large perturbations in model parameters. If the scatter plots doesn't show clearly any non-linear dependency, the PRCC provides good insight on global sensitivity, that is which parameters are most influential even if other parameters are simultaneously perturbed.
Of course there are other methods of assessing the global sensitivity, such as FAST.
For me PRCC is always the first step in assessing the global sensitivity of a model on the uncertainties in its parameters.
The local analysis (done mainly through calculating/approximating the derivatives with respect to the parameters) provides direct information on the effect of small parameter perturbations about their nominal values, but it does not indicate the effect of concurrent, large perturbations in all model parameters (in biomathematics that's almost always the case).
In order to calculate PRCC you simulate exactly those simultaneous large perturbations in model parameters. If the scatter plots doesn't show clearly any non-linear dependency, the PRCC provides good insight on global sensitivity, that is which parameters are most influential even if other parameters are simultaneously perturbed.
Of course there are other methods of assessing the global sensitivity, such as FAST.
Thanks, Jan! How would you compare the usefulness of PRCC to using Monte Carlo type methods (e.g. generating synthetic data sets using observed means and standard errors or using bootstrapping on the data) for assessing model sensitivity to parameter values?
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis