Statistical Modelling of Sit-to-Stand Movement in Normal and Parkinson's Disease Subjects

Abstract

At present, the area of modeling complex body movement patterns remains largely unexplored. The purpose of this research was applying sophisticated statistical models to a complex body movement in normal subjects and neurologically impaired individuals. Thirteen Parkinson’s disease (PD) subjects and twenty-nine normal controls were recruited to participate in a sit-to-stand motor task selected from basic functional movement activities. This study identified the rational function model as one, which could be used to describe the sit-to-stand task. The results showed that this model fitted the observed data well (R2 > 0.998). However, verification of the model showed that the residuals contained a sizeable autocorrelation factor. Results showed that PD subjects had a higher autocorrelation coefficient (mean = 0.91± 0.09) than the normal subjects (mean = 0.83± 0.12) for this task. The Cochrane-Orcutt method was applied to remove this factor and it also improved the fit of the model. Simulation was used to demonstrate the nature of the parameters in the model and significant differences in the model parameters were seen between normal and PD subjects. Such a new approach provides greater insight into body movement and may be of use in the development of therapeutic strategies to facilitate the performance of functional tasks by neurologically impaired individuals.