Estimating continuous and intermittent feedback models of postural control using the least squares method.

Abstract

Biomechanical models of postural control provide valuable insights into the mechanisms underlying balance. Although continuous and intermittent controller structures have been widely applied, their parameter identification from experimental data across diverse populations and sensory conditions remains underexplored. This study employs ankle torque signals and least squares (LS)-based methods to estimate and compare parameters of continuous and intermittent feedback models using sway data from young and older adults during quiet standing under varied sensory conditions. The LS, non-negative LS, and bounded-variable LS methods achieved high mean coefficients of determination ( $R^2>0.86$ ) for both models. Passive stiffness in the intermittent models remained consistent across sensory conditions, whereas active parameters varied for both models, reflecting adaptability. Simulations with these estimated models reproduced human sway patterns with reasonable accuracy. Despite some limitations, these techniques may contribute to advancing our understanding of postural control mechanisms.