The autoregressive time series modelling of stabilograms
Bräuer, D.; Seidel, H.
Acta Biologica et Medica Germanica 37(8): 1221-1227
1978
ISSN/ISBN: 0001-5318 PMID: 749458 Document Number: 131637
Power spectral density analysis of frontal and sagittal stabilograms in young male subjects with a normal vestibular function indicates the fitting of parametric time series models to stabilograms. Subjects were standing on a force platform under 3 different standing conditions. Stabilograms lasting for 2 min were processed by a TPAi computer with the CAMAC system. Following digital high-pass filtering, correlation functions and power spectral density distributions were estimated. Linear autoregressive models of increasing order up to 30 were fitted to stabilograms on the basis of the autocorrelation functions by means of a recursive scheme. The goodness of fit of the autoregressive models was significantly different in the standing conditions, planes and subjects. The orders of models for frontal stabilograms were higher than for sagittal ones, whereas the residual variances were lower than for sagittal stabilograms. Autoregressive modeling is a suitable approach for obtaining reliable spectral estimates and for characterizing the control system of body sway.