Prediction of Parkinson Disease Using Nonlinear Classifiers with Decision Tree Using Gait Dynamics

In the last decade, the second most common neurodegenerative disorder among the old people is the Parkinson's disease (PD). PD affects 2-3% of the population over the age of 65 years. Since the aging population rate is increasing at a faster rate all over the world, so many advanced techniques such as artificial intelligence and machine learning algorithms have been used to detect the progression of disease at early stage in short time. Few past research has been made to classify the PD from the healthy older peoples. They have used linear classification techniques for classification as well as linear dimensionality algorithm for feature selection. In this paper an attempt has been made to classify the PD group from the healthy control group by using nonlinear based classifier with decision tree and also nonlinear feature selection algorithm called Recursive Feature Elimination for selection of features. Finally, a performance comparison has been made between the original set of features as well as the reduced features. Our result does not able to find any significance difference in the performance metrics of the two feature sets, but achieved the classification accuracies ranging from 81.7% to 85.31%.