Performance Analysis of Different Classifiers for Tele-Diagnosis of Parkinson’s Disease

Parkinson’s disease (PD) is a second most progressive neurodegenerative disorder. Millions of people across the world are affected with this disease. In recent days, there are significant research has been reported for the screening of PD using Dysphonia features. In this study, a new weights generation method named as Kernel Fuzzy C-means Ratio based on different clustering technique (KFCM, FCM and KCM) has been proposed. The main aim of this work is to transform non-separable speech features in the dataset to a linearly separable such that the classification can be enhanced. In classification stage, six different classifiers are used to classify the weighted data and significant improvement in sensitivity, accuracy and specificity parameters are recorded.