Illustration of neural networks (NN). Each SNP occupies an input node and takes value 0, 1 or 2. The hidden nodes receive a weighted sum of inputs and apply an activation function to the sum. The output nodes then receive a weighted sum of the hidden nodes' outputs and, again, apply an activation function to the sum. For a 3-category classification (K = 3), three separate output nodes were used, with each node being specific to one category (low, medium or high). Classification was assigned to the category with the largest output value.