Standard deviation of individual parameters estimate


For a project, I am interested in both the population parameters and the individual parameters of my subjects. I am aware of the existence of the function icoef in order to retrieve the individual parameters from the fit, however, I was wondering if there was a way to get estimation errors for each of those individual parameters ? As I am going to need those as well.

Until now I used Pumas.FOCE() method to get my fit from my data, and I haven’t found a way to do that. And to be fair, I do not know/understand the FOCE method enough to know if that’s even a possibility (or if it would make sense).

For SAEM, we get a distribution for each individual parameter (or a sample from the distribution of the individual parameter) so I know that in this case it is normally possible to retrieve the estimated standard deviation for each individual parameter sample. However, I do not know how to get my hold on this information in Pumas.

So I was wondering if Pumas permits to retrieve such information or not? And if it is not yet possible, is that something which is considered in future releases?



hi @EmReg -
For any fit using Pumas.FOCE, you can get the variance-covariance matrix of the empirical Bayes estimates using Pumas.empirical_bayes_dist(myfit) function. This would give you a vector of length equal to the number of subjects and each element of the vector contains the variance-covariance matrix of that particular subject’s empirical Bayes estimates.


Thank you @vijay, that’s exactly what I needed!