where x i are vectors containing the measurements or features of gait data and y i are the corresponding class labels (i.e. +1=pathology, −1 =healthy). The SVM classifier is a linear hyperplane which separates two classes where the distance between the class boundaries to the hyperplane is the hyperplane margin. Data that is not linearly separable is first mapped via a nonlinear function ϕ:x⊂R n →R m n,m∈[1,∞) to a higher dimensioned feature space. The optimal hyperplane is the universal approximatorta