The measuring set and signal processing method for the characterization of human hand tremor

The paper presents an attempt to discriminate the most often observed three types of human tremor with objective techniques by measuring the acceleration of the hand and by calculating some characteristic features of these time series. Different mathematical descriptions have been adopted in order to generate the features. They have come from both second and higher-order statistics. A simple neural classifier has enabled the recognition of these three kinds of tremor with 3.6% error rate.