Background: Statistical characterization of tic behavior in Gilles de la Tourette syndrome (GTS) may provide insight into the dynamic functioning of the human central nervous system, as well as improve the quantitative assessment of tic symptom severity.
Methods: Twenty-two medication-free GTS subjects underwent videotaping of their tics. The intervals between temporally adjacent tics were measured, and the statistical properties of these intervals were assessed through graphical representation of frequency distributions, autoregressive integrated moving average (ARIMA) modeling, spectral analysis, and construction of first return maps.
Results: The frequency distribution of tic interval durations followed an inverse power law of temporal scaling. Spectral analyses similarly demonstrated that the spectral power density of tic interval duration scales inversely with frequency. ARIMA modeling suggested that the time series for tics are nonstationary as well as moving average processes. The first return maps demonstrated "burstlike" behavior and short-term periodicity in tics, and proved that successive tic intervals are not statistically independent. Graphic display of the time series confirmed shortterm periodicity, and in addition suggested the presence of period doubling.
Conclusions: These findings are suggestive though not conclusive evidence for the presence of a fractal, deterministic, and possibly chaotic process in the tic time series. These analytic methods provide insight into the temporal features of tics that commonly are described clinically (such as short-term bouts or bursting, and longer term waxing and waning), and they reveal certain important temporal features of tics that have not been clinically described. The methods may also prove useful in the improved characterization of tic symptom severity.