GMDH is a sorting-out of gradually complicated models generated from base function. The best model is indicated by the minimum of the external criterion characteristic. Multilayered procedure is equivalent to the Artificial Neural Network with polynomial activation function of neurons.
To forecast the values of future time steps of a sequence, specify the targets as the training sequences with values shifted by one time step. In other words, at each time step of the input sequence, the GMDH neural network learns to predict the value of the next time step.