Comparison of closed-loop system identification techniques to quantify multi-joint human balance control

More Info
expand_more

Abstract

The incidence of impaired balance control and falls increases with age and disease and has a significant impact on daily life. Detection of early-stage balance impairments is difficult as many intertwined mechanisms contribute to balance control. Current clinical balance tests are unable to quantify these underlying mechanisms, and it is therefore difficult to provide targeted interventions to prevent falling. System identification techniques in combination with external disturbances may provide a way to detect impairments of the underlying mechanisms. This is especially challenging when studying multi-joint coordination, i.e. the contribution of both the ankles and hips to balance control.

With model simulations we compared various existing non-parametric and parametric system identification techniques in combination with external disturbances and evaluated their performance. All methods are considered multi-segmental (both the ankles and the hips contribute to maintaining balance) closed-loop balance control. Validation of the techniques was based on the prediction of time series and frequency domain data. Parametric system identification could not be applied in a straightforward manner in human balance control due to assumed model structure and biological noise in the system. Although the time series were estimated reliably, the dynamics in the frequency domain were not correctly estimated. Non-parametric system identification techniques did estimate the underlying dynamics of balance control reliably in both time and frequency domain. The choice of the external disturbance signal is a trade-off between frequency resolution and measurement time and thus depends on the specific research question and the studied population.

With this overview of the applicability as well as the (dis)advantages of the various system identification techniques, we can work toward the application of system identification techniques in a clinical setting.