model predictions for spatially corrected logistic model

Hi

Does any one have an idea how you make a model prediction for a model that is corrected for spatial autocorrelation? I used the cote-corrected covariate mean in the formation of new data. Is it the normal way you predict a model. I tried doing this but a lot my data gets removed and with the little data it is not able to plot.

Some advise or suggestions would be great.