Estimation of UPDRS Finger Tapping Score by using Artificial Neural Network for Quantitative Diagnosis of Parkinson's disease

The purpose of this study was to estimate UPDRS finger tapping score with Parkinson's disease patients by using an artificial neural network. The measurement system was composed of a pair of 3-axis accelerometers, a pair of touch sensors, an analog to digital converter and a personal computer. Firstly, the accelerations during the finger tapping were measured with 44 normal subjects and 17 Parkinson's diseases subjects by using this system. Secondly, the four features were extracted from the obtained accelerations. Finally, the UPDRS finger tapping score was estimated by using a three-layer artificial neural network model.