I am working on a ‘quick’ solution to allow GLMMs to be used in a DataSHIELD setting. So far I have been adapting the ds.glmSLMA function (and corresponding server side functions) to give me something to start from. I now have the basic mechanics of this working such that I can fit a model on the server side to each dataset and bring the results back for meta analysis.
The challenge now is to identify any potential disclosure traps for GLMMs. Any thoughts would be welcome!