Identification and classification of neuro-degenerative diseases using feature selection through PCA-LD

Neuro-degenerative diseases have become a main cause for the disturbance in the walking pattern of people. Problems that are entirely related to the gait affect not only the walking pattern but also causes, many problems including lack of social activities, health related issues including both the physical and mental problems and much more. Thus the main aim of the paper is to identify and classify the patients suffering from various neuro-degenerative diseases from the human gait. Four categories have been used including (Amyotrophic laterals sclerosis, Huntington, Parkinson's, Control subject). Seven parameters have been used for the proposed work which include, left stride interval, right stride interval, left swing interval, right swing interval, left stance interval, right stance interval and double support. As the number of samples has been found to be more thus features have been extracted to reduce the computation and eliminate the redundancy. Eight features have been used for all the seven parameters namely mean, standard deviation, variance, skewness, kurtosis, energy, entropy, correlation. Among all the 56 features and parameters some of the parameters have been found to be redundant hence feature selection has been done using PCA-LDA in order to improve performance as well as reduce the dimensionality. Algorithm has been found to work best in case of HD category with the accuracy of 93% and reduction of dimensions from 56 to 3. Similarly the accuracy of PD category has been found to 89% with the reduction from 56 to 20 dimensions. Algorithm has been applied for all the neuro-degenerative disease and the accuracy has been found to be 85%.