Correlation among joint motions allows classification of parkinsonian versus normal 3-D reaching

In this paper, an objective assessment for determining whether a person has Parkinson disease is proposed. This is achieved by analyzing the correlation between joint movements, since Parkinsonian patients often have trouble coordinating different joints in a movement. Thus, the auto-correlation coefficient of single joint movements and the cross-correlation between movements in a pair of joints (hand, wrist, elbow, and shoulder) were studied. These features were used to train and provide classification of subjects as having or not having Parkinson's disease using the least square support vector machine (LS-SVM). Experimental results showed that using either auto-correlation or cross-correlation features for classification provided over 91% correct classification. Using both features together provided better performance (96.0%) than using either feature alone. In addition, the performance of LS-SVM is better than that of self-organizing map (SOM) and k-nearest neighbor (KNN) in this case.