Train NARX-Neural Network multiple Inputs/Outputs data from multiple experiments

I have a system with 3 inputs and 4 outputs, the data is stored in 3D matrices as follows:

Input: X(i,j,k) = [400 x 3 x 10]

Output: Y(i,j,k) = [400 x 4 x 10]

where i: number of measurement point.

j: number of variable (input or output variable).

k: number of experimental run.

I need to pass this data through the command 'train' so I obtain the NN parameters.

How can I manage the data to achieve it? Which commands should I use? Could you give me a sequence of commands to process the matrixes so I obtain an arrange able to be place in the command 'train'?

Thanks in advance!