Correlation of Kinematic Parameters and UPDRS Scoring during Finger Tapping movement in Parkinson's Disease

In Parkinson's disease, the clinical assessment of the upper-limb motor test severity occurs through the UPDRS (Unified Parkinson's Disease Rating Scale). With the perspective of building a system for at-home monitoring of Parkinson's patient status, we investigate algorithms for an automatic scoring of the UPDRS upper-limb motor tests. Building on a previuous work, where we developed a system for the automatic evaluation of kinematic parameters relevant for Parkinson's desease, we adopt a machine learning approach to the automatic scoring problem. A pool of Parkinson patients have been scored by neurologists for UPDRS upper-limb motor tests and the kinematic parameters of their performances acquired. The most significant kinematic parameters are identified by a PCA analysis. Different classifiers have been trained with the neurologists UPDRS scores and the associated kinematic parameters of the patients. The results of the investigation are presented.