How to predict or estimate output vector when the input is a matrix instead of vector using Nonlinear regression techniques

Hi, I have been working on a prediction problem from last few weeks. I have some experimental data of 70 sets, Lets say, X(70x5) is the input matrix & Y(70x1) is the output vector. So for a linear model, we can solve it by taking Y = X*W, (W is a 5x1 vector) and simply with W== mldivide((X'*X),X'*Y) But How to solve this problem using some Nonlinear techniques. Almost all the predefined functions like Nlinfit are accepting only vectors as the inputs. How to solve it when the output is depending upon multiple inputs. (i.e. input is a matrix) My objective is to build a model first and then to use it for prediction. Thank you in advance.

[Wangyan Li]

have you found solotion yet?