A video game based hand grip system for measuring muscle force in children

Abstract

Background: While new therapies are continuously introduced to treat muscular dystrophy, current assessment tests are challenging to quantify, cannot be used in non-ambulatory patients, or can de-motivate pediatric patients. We developed a simple, engaging, upper-limb assessment tool that measures muscle strength and fatigue in children, including children with muscular dystrophy. The device is a bio-feedback grip sensor that motivates children to complete maximal and fatiguing grip protocols through a game-based interface.

Methods: To determine if the new system provided the same maximum grip force as what is reported in the literature, data was collected from 311 participants without muscle disease (186 M, 125 F), ages 6 to 30, each of whom played the four minute grip game once. We compared maximum voluntary contraction at the start of the test to normative values reported in the literature using Welch's unequal variances t-tests. In addition, we collected data on a small number of participants with muscle disease to determine if the assessment system could be used by the target patient population.

Results: Of the 311 participants without muscle disease that started the test, all but one completed the game. The maximum voluntary contraction data, when categorized by age, matched literature values for hand grip force within an acceptable range. Grip forced increased with age and differed by gender, and most participants exhibited fatigue during the game, including a degradation in tracking ability as the game progressed. Of the 13 participants with muscle disease, all but one completed the game.

Conclusions: The study demonstrated the technical feasibility and validity of the new hand grip device, and indicated that the device can be used to assess muscle force and fatigue in longitudinal studies of children with muscular dystrophy.

Keywords: Game-playing; Muscle fatigue; Muscle force assessment; Muscular dystrophy.

Fig. 1

a Hand-grip dynamometer and associated electronics interface box. b System in use at the 2018 Minnesota State Fair

Fig. 2

a Screen shot of the calibration task from the Rocket Launch application. b Screen shot of the tracking task from the Rocket Launch application

Fig. 3

Task interval time line used in the Rocket Launch game

Fig. 4

Example of run data from a 9 year old male participant. Target is percent MVC0. MVC0, MVC1 and MVC2 indicate the three MVC tests during the game

Fig. 5

MVC0 (mean+95%CI) of male participants ($\mathit{n}=139$, dark bars) compared to normative male data (light bars) from multiple sources [, , –54], grouped by 2 year age ranges from 6 to 19 and 5 year age ranges from 20-34. Significant differences marked by *($\text{p}<.05$) and **($\text{p}<.01$)

Fig. 6

MVC0 (mean+95%CI) of female participants ($\mathit{n}=98$, dark bars) and normative female data (light bars) from multiple sources [, , –54], grouped by 2 year age ranges from 6-19 and 5 year age ranges from 20 to 34. Significant differences marked by *(p <. 05) and ** ($\text{p}<.01$)

Fig. 7

Grip strength by age in a healthy population of males ($\mathit{n}=139$) and females ($\mathit{n}=98$) with trendlines for male ($r^2=0.67$) and female ($r^2=0.46$). The trendlines show that MVC increases with age for both genders

Fig. 8

MVC % (mean+SD) during MVC0 ($\text{M}=100.0$, $\text{SD}=0.0$), MVC1 ($\text{M}=80.2$, $\text{SD}=12.6$) and MVC2 ($\text{M}=76.2$, $\text{SD}=13.7$) periods of gameplay in healthy population ($\text{N}=237$ for each group). There was a significant difference between groups, $\text{F}(2,708)=335.8$, $\text{p}<.001$ with post hoc comparisons using Bonferroni correction indicated that all three groups were significantly different from each other (***$\text{p}<0.001$)

Fig. 9

RMS tracking error over the entire game by MVC0 in a healthy population ($\text{N}=237$). The two variables are not strongly correlated, $\text{r}(235)=-0.22$, $\text{p}<0.001$

Fig. 10

RMS tracking error over the entire game by age in a healthy population ($\text{N}=237$). The two variables are not strongly correlated, $\text{r}(235)=-0.22$, $\text{p}<0.001$

Fig. 11

Compares RMS tracking error (mean + SD) for the 1st and 2nd 40, 60 and 80% of MVC0 target periods in a non-impaired population ($\text{N}=237$ in each group). For the 40% MVC0 target, tracking was better in the second half of the test while for the 60 and 80% MVC0 targets, tracking was worse in the second half

Fig. 12

Example of run data from a 20 year old male participant with Duchenne muscular dystrophy