How do I use the function generated from the Neural Network toolbox?

I'm trying to learn how to use the NN toolbox, but i can't really understand what are the inputs and outputs for the function generated after the training of my model.

I'm using the time series example of pollution mortality, which tries to predict 3 types of mortality from 8 different types of data. The function tries to explain the arguments needed:

% [y1,xf1,xf2] = myNeuralNetworkFunction(x1,x2,xi1,xi2) takes these arguments:
%   x = 8xTS matrix, input #1
%   1 = 3xTS matrix, input #2
%   xi1 = 8x1 matrix, initial 1 delay states for input #1.
%   xi2 = 3x1 matrix, initial 1 delay states for input #2.
% and returns:
%   y = 3xTS matrix, output #1
%   xf1 = 8x1 matrix, final 1 delay states for input #1.
%   xf2 = 3x1 matrix, final 1 delay states for input #2.
% where TS is the number of timesteps.

I thought that i only needed to provide the next 8 inputs so that the function would give me the 3 outputs predicted, so what is the need for so many inputs and outputs? Also, what exactly is the "number of delays"? I chose 1 because I only want to predict the next result for the timeseries, but i'm not sure if that's right.

Thanks in advance

[imene. B]

To people wondering how this is done, i guess its a traiend NARX network used to generate a code from getFunction:

genFunction(net,'the name of the file you will save your function in it');

some helping pages:

https://www.mathworks.com/help/deeplearning/ref/network.genfunction.html

https://www.mathworks.com/help/deeplearning/ug/deploy-shallow-neural-network-functions-and-objects.html

https://www.mathworks.com/matlabcentral/answers/355286-how-to-give-multiple-inputs-to-the-train-function-of-neural-network