Use a spatial lag model to predict (in R)

I have two polygon datasets, one for year 2010 and another for year 2015, each containing multiple features with a number of different variables.

I used a spatial lag model to predict the dependent variables (urban density) in the 2010 dataset. This was done in R.

For the 2015 data I only have explanatory variables, no urban density, and would like to predict urban density for 2015 using the results/coefficients from the 2010 spatial lag model.

I can do this with a linear regression model in R by using the predict function, see link. But I can’t get that to work on spatial data, wont work because it doesn’t take space in to consideration.