Quantifying and defining criteria for balance proficiency for a specific balance task is an important factor when improving one’s balance. Motion-based games are selected to emphasize 5 distinct balance tasks as a method of improving balance and reducing fall risk in individuals’ post-stroke. Prior research in the area of game-based balance training has shown that motion-based balance tasks can improve balance and stability in post-stroke individuals. Variables step length and time in single-leg stance (SLS) were analyzed to identify parameters for balance proficiency in 2 specific balance tasks.
Research in step length and time in SLS as a method to quantify balance have also been conducted, though primarily for slip-based balance training. This research concluded that decreasing step length while increasing gait speed can improve stability.
Nine post-stroke participants took part in 5 video game balance tasks using Microsoft Xbox Kinect with several different surfaces. Balance tasks were organized into 5 categories: time in SLS/kicking, anterior/posterior stepping, medial/lateral (ML) stepping, feet in place, and trunk rotations with feet in place. Participants played about 5 minutes per task during each session for a total of 10 sessions, with the first session introducing the participants to the specific tasks. The second and last sessions used Cortex Motion Capture software (Motion Analysis Corp) with 8-cameras at 120 Hz and the 2016 Helen Hayes 37-marker set to record participants’ kinematic and center of mass data. MATLAB code was created for custom analysis and graphing of kinematic data not provided by the Cortex software.
I chose 2 categories of the 5 balance task groups to analyze: medial-lateral stepping and time in SLS, both on a flat surface. The medial-lateral stepping task involved the participant using both feet to step side to side quickly in an inconsistent pattern. The time in SLS balance task involved quick kicks with each foot, alternating between the left and right foot. Within the 2 categories, I analyzed XYZ resultant step length in the ML stepping task and length of time in SLS in the SLS/kicking task. The participants with the highest average step length and SLS time for the tasks were used. Cortex Motion Capture doesn’t calculate these variables directly, so MATLAB code was created which monitored the heel and toe markers x, y, and z positions to determine whether a given foot lifted off the surface during the task in order to calculate the 2 variables.
RESULTS AND DISCUSSION
Figures 1, 2, along with Table 1 summarize data for the ML stepping and time in SLS balance tasks for a single participant.
The step length data (Figure 1) (Table 1) shows a higher average step length with the left foot than that of the right foot, along with more steps taken. The SLS results (Figure 2) (Table 1) display a greater number of kicks with the left foot, but a higher mean time in SLS with the right foot in less steps.
Through observing participants in both balance tasks over repeated sessions, balance proficiency criteria for the 2 activities can be identified. To have proficient balance at the ML stepping task, the participant must be able to take wider steps from side to side. To demonstrate balance proficiency with the SLS/kicking task, the participant must be able to kick their foot out quickly at regular intervals. The ability to perform this multiple times and with both feet consistently would be an example of proficient balance for that task. These 2 ways to quantify balance vary a great deal due to differences in the many types of balance tasks.
After learning the specific balance training protocol, the Cortex data collection software, and the existing MATLAB codebase, I was able to identify what data analysis scripts and functions would need to be created for the specific variables analyzed. Creating these analysis programs allowed us to quantify balance improvement related variables such as the resultant step length and time in SLS.