Statistical feature analysis of tremor signal

Tremor is an unintentional oscillatory movement of body parts. It is observed in healthy people as well as in people with disorders. While the symptoms are similar, tremor signals have different frequencies and amplitudes. The method we proposed is based on higher order statistics. Power spectrum and bispectrum are two main features of higher order statistics. We estimated the bispectrum and power spectrum of tremor signals and compared these features for each subject with and without medication. It was observed that higher bispectrum and power spectrum correspond to cases without medication. We also generated an efficient mechanism for performing an analysis on the distribution of the power spectrum and bispectrum of tremor signal to examine the influences of gender, age and the progress of disease under a certain medication condition and evaluate the success of a treatment.