The primary objective of this study was to use linear stochastic modeling approach to investigate parameters which may be sensitive enough to detect and quantify the changes in postural instability (PI) related to the progression in Parkinson’s disease (PD). Data collected in a previous study were analyzed in the current study. Participants with mild PD (n=13), moderate PD (n=10) and age range match healthy controls (HC, n=21) were instructed to stand in a comfortable self-selected natural stance on a force platform in both eyes open (EO) and eyes closed (EC) conditions. The foot-floor reaction forces were used to calculate the center of pressure (COP) time series. This COP time series was fitted by two different linear stochastic models:​ 1) an autoregressive (AR), and 2) an autoregressive moving average (ARMA) model. The postural control system was modeled as an inverted pendulum to describe pure body mechanics and a proportional, derivative and integral (PID) strategy was assumed for balance regulation. Swiftness, damping and stiffness parameters were extracted from the AR model. Natural frequency and damping ratio were extracted from the ARMA model.

The statistical analysis (ANOVA) of these parameters revealed significant differences in stiffness and swiftness parameters between the HC and moderate PD population in the EO condition. These three parameters showed trends with progression of PD. The swiftness parameter showed decreasing mean values as PD severity increased, indicating that PD caused slower reactions to small deviations from equilibrium when compared to healthy controls. The mild and moderate PD, compared to HC, demonstrated by higher mean values of stiffness, suggesting a more rigid control strategy. The analysis of damping parameter revealed that the PD, compared to HC, may have a reduced ability to attenuate sway velocity during quiet stance as indicated by lower mean values of damping parameter and damping ratio. The natural frequency did not show significant trends in EO condition, but revealed an increasing trend with progression of PD. This could indicate that the PD could have larger number of deviations of COP from equilibrium. The analysis of effect of condition (EO, EC) revealed significant differences in all the five parameters. The stiffness, damping parameter and damping ratio had higher mean values for EO, compared to the EC condition, indicating the vital role that the visual feedback plays in detecting small perturbations from equilibrium leading to a better posture regulation in EO condition. The swiftness parameter and natural frequency indicated higher mean values in EC, compared to the EO condition, suggesting that the various sensory cues might be weighted differently in EO and EC conditions.

Future studies should investigate the sensitivity of these calculated parameters to changes in PI in PD using a larger sample size and longer duration of trials. Also the variations in these parameters in response to dynamic tasks such as gait initiation and balance recovery should be considered in future studies.