nlmefit(*) support for nonlinear marginal model?

Dear Mathworks, I'm unsure from the documentation if the nlmefit() function can support a nonlinear marginal model of the form: y_i = f(Xi,\beta) + Zi*bi + \epsilon_i. That is, the random effects bi enter the model linearly. It seems from the documentation that random effects can only be directly added to a corresponding fixed effect, e.g. for fixed effect variable A, becomes (A + a) when adding a random effect, and is directly inserted into function f. Note that for the model I give, random effects may exist without being directly added to a particular FE. Can this be done in nlmefit? Thanks!