Chaos 5:52-56.
This paper describes our analysis procedure for long-term tremor EMG recordings, as well as three examples of applications. The description of the method focusses on how characteristics of the tremor (e.g. frequency, intensity, agonist-antagonist interaction) can be defined and calculated based on surface EMG data. The resulting quantitative characteristics are called ``tremor parameters''. We discuss sinusoidally modulated, band-limited white noise as a model for pathological tremor-EMG, and show how the basic parameters can be extracted from this class of signals. The method is then applied to (1) estimate tremor severity in clinical studies, (2) quantify agonist-antagonist interaction, and (3) investigate the variations of the tremor parameters using simple methods from time-series analysis.