Anomaly gait classification of Parkinson disease based on ANN

Abstract

The aim of this study is to investigate the potential of Artificial Neural Network (ANN) as classifier for distinguishing gait pattern between normal healthy subjects and Parkinson Disease (PD) patients. Since it has been proven by various researchers that PD patients owned significant gait deviation compared to normal adults, hence this study are conducted and will mainly focused on the basic, kinetic and kinematic measurements of human gait. Initial findings attained confirm that the ANN classifier successfully distinguished gait pattern between normal and PD gait with 81.25%, 81.25% and 84.38% success rate respectively for basic, kinetic and kinematic features solely. In addition, data fusion is performed for both basic and kinetic features, followed by basic and kinematic, kinetic and kinematic and all the three features. It was found that results of accuracy has increased to 87.5% based on data fusion of two or more features.