Changes in posture and balance performance during five days of wobble board training

Abstract

Introduction: Human postural control is facilitated through postural movements such as ankle-, hip-, or multi-joint strategies. Principal component analysis (PCA) applied to kinematic marker offers a novel approach study the structure of postural movements by identifying and quantifying correlated segment motion. This study investigated if the structure of the postural movements changes as subjects learn to master a balance task (standing on a wobble board). It was hypothesized that the relative contribution of principal components quantifying the main types of body sway (e.g. ankle strategy) to the whole postural movements would decrease as subjects improved in performance, while the contribution of higher-order movement components would increase. Methods: Eleven healthy male volunteers (age 25.1±1.7, weight 77.2±5.8 kg, height 1.80m±0.07) conducted a total of 25 120-second quiet stance trials on a wobble board, 5 trials per day during 5 consecutive days. The subjects’ postural movements were recorded with a standard 3D-camera system (ProReflex, Qualisys INC., Gothenburg, Sweden) using 49 reflective markers distributed over all major body segments. For each timeframe, a 147-dimensional posture vector was defined that included all marker coordinates. The posture vectors of all trials of a subject were normalized and assembled into an input matrix for the PCA. The structure of a subject’s postural movements was then characterized by calculating the relative contribution (RC) of the first 10 principal movement components (PCs) to the entire postural variation in one trial. For each trial, a “balance score” was calculated by totalling the standard deviation of the vertical position of 4 markers placed on the wobble board. A repeated measures ANOVA (Sidak correction) was conducted to determine differences in RC or balance score between trials. Results: Balance performance on the wobble board improved over the first 2-3 test days with the balance score decreasing from 52.1±11.0 in the first trial to levels below 34.2±6.1 in all trials of the 4th and 5th day (mean±stdev). This change was significant (F(1,24)=11.17, p<0.007). However, no systematic changes were observed in the structure of the postural movements as quantified by the first 10 PCs: F(1,24)<1.42, p>0.98 in the RC calculated for the first 10 PCs. Conclusion: The hypothesis was not confirmed. The results of this study suggest that the improvement in performing the wobble-board balance task was not related to changes in the structure or organization of postural movements as quantified by PCA-RC.