A Robust Approach for Parkinson Disease Detection from Voice Signal

Abstract

Parkinson’s disease (PD) is a common brain disorder that is associated with slow speech and difficulty with articulation. Mainly, clinical experts analyse patient’s voice to detect PD. In this paper, we proposed a robust model using empirical mode decomposition (EMD) and machine learning algorithms. Firstly, the acquired voice signals are pre-processed and segmented into small intervals. Then, each segment is sent to EMD model. A set of entropy features are extracted and then they are fed into a K-nearest neighbor (KNN), least squares support vector machine (LS-SVM), bagged tree, SVM (support vector machine), and K-means. The proposed model is evaluated using a publicly available. Our findings showed that the proposed framework can classify voice signals with a 97% accuracy.