Classification of autism children gait patterns using Neural Network and Support Vector Machine

In this study, we deemed further to evaluate the performance of Neural Network (NN) and Support Vector Machine (SVM) in classifying the gait patterns between autism and normal children. Firstly, temporal spatial, kinetic and kinematic gait parameters of forty four subjects namely thirty two normal subjects and twelve autism children are acquired. Next, these three category gait parameters acted as inputs to both classifiers. Results showed that fusion of temporal spatial and kinematic contributed the highest accuracy rate for NN classifier specifically 95% whilst SVM with polynomial as kernel, 95% accuracy rate is contributed by fusion of all gait parameters as inputs to the classifier. In addition, the classifiers performance is validated by computing both value of sensitivity and specificity. With SVM using polynomial as kernel, sensitivity attained is 100% indicated that the classifier's ability to perfectly discriminate normal subjects from autism subjects whilst 85% specificity showed that SVM is able to identify autism subjects as autism based on their gait patterns at 85% rate.