Analysis of the body sway in the upright standing posture fitting autoregressive and feedback models

Abstract

Body sway was analyzed by autoregressive and feedback models in an attempt to clarify the mechanisms involved in posture regulation.Sway of the head, shoulder, hip, knee and foot during upright standing were recorded polygraphically and stored in a PDP-11 computer. The data were processed by a standard method and 5-dimensional autoregressive and feedback models in a specially designed program. The results were displayed on a cathode ray tube and teletype as determinant of noise correlation matrix, power spectrum, correlogram, transfer function, impulse response and relative power contribution. Forward-backward (FB) and leftright (LR) sways during upright standing were examined in normal subjects with eyes open and closed.1. The results of the standard method had advantage of showing actual body sways, but the gain diagrams of the power spectrum were not smooth. They were very smooth with the 5-dimensional autoregressive and feedback models. Furthermore, since the influence of past sways and feedback from five parts of the body were taken into account in the autoregressive model and feedback model, respectively, the resulting data had additional implica-tions. There is a feedback mechanism in posture regulation. The feedback model is superior for analysis of body sway in the upright standing posture.2. The results obtained from the 5-dimensional feedback model in normal subjects were:a) The determinant of noise correlation matrix was close to 1. It indicated that the system composed of sways of the head, shoulder, hip, knee and foot had good feedback. The composition of feed-back was better for FB sway than for LR sway.b) The power spectrum of the head, shoulder, hip and knee were high at 0.1-0.2Hz and decreased thereafter as the frequency range rose from 0.1 to 10 Hz.c) The correlogram with the eyes open revealed a Markov process, while the correlogram with the eyes closed indicated a periodic component of about 0.2 Hz.d) In the transfer function, sways of the head, shoulder, hip and knee with small amplitude were demonstrated as output.e) The impulse response converged immediately. Recovery of posture against external disturbances was satisfactory.f) The relative power contribution indicated that each part of the body was not appreciably influenced by other parts but showed its intrinsic sway in a high frequency range from 0.2 to 1 Hz.