The Use of a Low-Cost Gaming Platform as a Substitute to Medical-Grade Force Plate in Objective Evaluation of Balance

Abstract

Posturography is a tool for quantitative measurement of balance in various static and dynamic test conditions. Current medical grade Posturography systems are very costly and rare outside research institutes. The Wii Balance Board is an inexpensive gaming device which works on similar principles and uses the center of pressure of an individual as an input for motion-controlled gaming. The aim of this study was to assess the validity of using a gaming platform (Wii Balance Board) as a substitute to medical grade Posturography machine. The objectives were to compare the measures of postural sway obtained from a gaming platform with those from a medical grade Posturography machine and to demonstrate a low-cost method of analysis of balance using the gaming platform. Descriptive Observational study. Individuals aged 18–65 years and having no apparent dysfunction of balance were assessed objectively using Wii Balance Board and Medical-grade Posturography machine. A program was developed in Scilab for computation of sway parameters. The results were compared in terms of correlation coefficient, Bland-Altmann plots, and ability to detect abnormal sway. A Pearson’s correlation coefficient ranging from 0.53–0.91 was observed in various test conditions, with an overall value of 0.62 between the results of the two machines. The sensitivity was 69.39%, specificity was 73.16%, positive predictive value 24.29%, negative predictive value 95.07%, and accuracy 72.75%. Thus, we can conclude that the Wii Balance Board can be used as a low-cost substitute to medical grade Posturography machine for quantification of balance in situations where precise measurement of balance is not required.

Introduction

There are various methods for evaluation of balance which can be classified as being either subjective or objective. The clinical tests commonly performed are all subjective and do not provide any quantification of balance of the individual. Posturography is an objective test of evaluation of balance which is capable of providing quantitative data about one’s stability. Balance tests using Posturography can be static or dynamic, depending on the machine capabilities. Static Posturography is done using a pressure sensitive force plate which has pressure sensitive gauges in each of its quadrants. The forces exerted on the gauges are used to compute the coordinates of center of pressure, which are then used to calculate the postural sway of the individual standing on the Posturography machine [1, 2]. There are various medical grade force plates available which have been used in research for balance. However, the costs are often prohibitive, they lack portability and availability of these machines continues to be rare outside the research institutes [3, 4].

Motion controlled gaming utilizes hardware to track the position of the human player to control the characters in the game. One such technology involves tracking of centre of pressure of the player using a gaming platform. These low-cost accessories connect wirelessly to the gaming console and work on a principle similar to medical grade force plates [4]. The Nintendo Wii Balance Board (WBB) (Nintendo Co. Ltd, Kyoto, Japan) is one such gaming platform that has been subject to much of balance related research by the neurologists and neuro-otologists [3]. It has been used for research involving objective evaluation of balance and neurological rehabilitation owing to its lower cost, ease of use, portability, availability of extensive documentation and various open-source Posturography software [4, 5]. However, there have been mixed reports about the validity of the results obtained from device, and its ability to detect abnormal balance as compared to the medical-grade Posturography machines.

Rey Martinez J et al. compared stabiliometric measurements from the WBB to a clinical grade Posturography machine. They developed an open-source software for performing static balance tests using the WBB on macOS. They found a strong positive correlation between the results obtained from both machines and thus were able to validate the use of their software and the WBB for Posturography [6]. Clark RA et al. systematically reviewed the reliability and validity of WBB for evaluation of static balance. They identified 21 articles assessing its validity for Posturography, out of which 12 articles found a strong validity of WBB. The remainder studies showing poor to moderate validity had low quality ratings due to their limitations. They concluded that the results obtained from WBB are valid as those from commercial force platforms, but owing to its technical limitations, it should not be considered a replacement for medical grade Posturography machines [5]. Despite much available research literature, none mentions the sensitivity, specificity, positive and negative predictive values of identifying abnormal postural sway using a gaming platform. Hence a dilemma still exists regarding detection of abnormal balance using WBB for performing Posturography in balance clinics.

The aim of the current study was to determine if a gaming platform may be used as a substitute for a standard medical grade Posturography machine. The objectives were to compare the measures of postural sway obtained from the gaming platform with those from a medical grade Posturography machine and to develop and demonstrate a low-cost method of analysis of balance using the gaming platform.

Methods

The study was conducted at the Otorhinolaryngology department of a tertiary care centre from January to March 2019. The inclusion criteria included healthy persons of either gender in the age group of 18–65 years without any apparent balance dysfunction. The exclusion criteria included persons with illnesses or injuries affecting satisfactory performance of balance tests, or persons in whom performing Posturography involved risk of fall due to their existing illnesses. Based on the data from the previous studies, a sample size of 78 was determined to test the null hypothesis that there is no difference in between the mean values measured from the two devices [3]. A total of 115 persons enrolled in the study, out of which four persons were excluded based on the exclusion criteria. Hence a total of 111 persons meeting the inclusion criteria were included in study.

The patients were counselled about the research protocol and informed consent was obtained regarding their participation in the current study. NeuroCom® Basic Balance Master® (BBM) (NeuroCom International Inc., Clackamas, USA) was used as the medical grade Posturography machine and was taken as the standard in the current study. Wii Balance Board (WBB) was the gaming platform used as the research tool of the current study. Computerised Static Posturography (Fig. 1a, b) using modified Clinical Test of Sensory Interaction on Balance (mCTSIB) was done on both of the platforms under four standard test conditions [7]. Condition I involved the person standing on the force plate with Eyes open. Condition II involved standing on the force plate with Eyes closed. For conditions III & IV, a foam block (provided along with BBM) was placed on top of the force plates. Condition III involved standing on the force plate with foam block and Eyes open. Condition IV involved standing on the force plate with foam block and Eyes closed. A recording time of 30 s was used in each of the test conditions.

Fig. 1.

a Posturography with BBM, b Posturography with WBB, c BBM & WBB devices, d BrainBloX software

The tests were conducted first with the medical grade Posturography machine (BBM) and followed by the gaming platform (WBB) for the same person in the same environment (Fig. 1c). The BBM was connected using its data wire cable to a Laptop Computer running Microsoft Windows 7® (Microsoft Corp. Redmond, WA). Data of postural sway was captured from BBM by the software provided with it by the manufacturer. The WBB being a wireless device, was connected to the Laptop Computer by pairing it over Bluetooth. The “CU BrainBLoX” software (Neuromechanics Laboratory, University of Colorado Boulder) was used to plot the centre of pressure and record the postural sway from WBB (Fig. 1d) [8]. From both the machines (BBM & WBB), the raw data was saved as data files, separately for each of the four test conditions for every patient.

The raw data file exported from the BBM is in the form of a text file containing the time elapsed (in 1/100th of a second), data of forces recorded from the four gauges in each of the four quadrants of the Posturography platform, along with the calculated values of Centre of Force along the x- and y-axis. For the WBB, the CU BrainBLoX software accomplishes the similar task, recording the postural sway, and exporting the raw data as a text file containing the time elapsed (in milliseconds) and the forces recorded from the four gauges in each of the four quadrants of the WBB [8]. The raw data from both BBM & WBB is sampled at about 100 times a second and is stored as separate rows in the data files. The data files were saved separately for each of the four test conditions for every patient.

Further analysis and mathematical calculation of sway parameters described in literature was done using a self-written computer program in a free and open-source numerical computation software “Scilab” (https://www.scilab.org/). The program was designed to open a data file, read its contents, and perform further mathematical calculations on the raw data. First, the Centre of Pressure along the x-and y-axis (COP_x & COP_y) were calculated for all the data points of a trial, as per the formulae described in previous studies [9, 10].

$$COPx=\frac{X}{2}\ast \frac{\left(TR+BR\right)-\left(TL+BL\right)}{TR+BR+TL+BL}$$

$$COPy=\frac{Y}{2}\ast \frac{\left(TR+TL\right)-\left(BR+BL\right)}{TR+BR+TL+BL}$$

where X and Y represent the length and width of the force platform respectively, and TL, TR, BL, BR represent the forces recorded by its top left, top right, bottom left and bottom right quadrant force gauges. The calculated values of COP_x and COP_y for all data points of the trial was then used for further calculation of sway parameters (outcome variables) namely the path length or excursion (PL), sway velocity (V), sway area (SA), 95% Confidence Ellipse Area (CEA) and the Equilibrium Score (ES) for the CEA as per the formulae described in previous studies [11–13].

$$PL=\sum _{n=1}^{N-1}\sqrt{{\left(A{P}_{n+1}-A{P}_n\right)}^2+{\left(M{L}_{n+1}-M{L}_n\right)}^2}$$

$$V=\frac{\mathit{PL}}{T}$$

$$SA=\frac{1}{2T}\ast \sum _{n=1}^{N-1}\left|A{P}_{n+1}\ast M{L}_n-A{P}_n\ast M{L}_{n+1}\right|$$

$$CEA=2\pi F_{0.05\left[2,N-2\right]}\sqrt{(S_{\mathit{AP}}^2\ast S_{\mathit{ML}}^2-S_{\mathit{APML}}^2)}$$

$$E{S}_c=\frac{P_{95}CE{A}_c-\left(CE{A}_c-P_{5}CE{A}_c\right)}{P_{95}CE{A}_c}$$

where N represents the number of data points recorded, ML and AP represent the COP_x and COP_y data points respectively, T denotes the time duration of each trial of Posturography, S_ML and S_AP represent the standard deviations of the COP_x and COP_y time series, S_APML represents the covariance of COP_x and COP_y time series, F_0.5[2,N-2] is the F-statistic at 95% confidence level with N data points, c denotes the test conditions (I-IV), P₅CEA_c and P₉₅CEA_c represent the 5th and 95th percentile of CEA for each test condition.

To further provide a visual plot of postural sway of the individual, the program developed in Scilab was also programmed to plot the stabiliogram (graphical plot of movement of centre of pressure with time, Fig. 2a) and the statokinesiogram (graphical plot of centre of pressure points in the antero-posterior and the lateral direction, Fig. 2b). These can be used to review the postural sway of the individual in two dimensions.

Fig. 2.

a Stabiliogram, b Statokinesiogram plotted from Scilab

The above sway parameters computed from the BBM & WBB were then tabulated and stored in a Microsoft Excel® spreadsheet for further comparison and data analysis. IBM SPSS Statistics® version 20 (IBM Corp. Armonk, NY) was used for statistical tests. One Sample T-test, Linear regression, Pearson’s correlation coefficient and Bland Altmann plots were used to compare the results obtained from the two machines. One Sample T-test was used to determine if there was a significant difference in the measurement from the two devices. Correlation coefficient was interpreted as described by Mukaka MM [14]. The number of individuals having postural sway values beyond the reference normal values for the two machines was counted and was used for calculation of sensitivity, specificity, positive and negative predictive values, and the accuracy.

Results

A descriptive observational study was conducted at a tertiary care centre from January to March 2019. A total of 115 persons enrolled in the study, out of which four persons were excluded as per exclusion criteria. Hence a total of 111 persons meeting the inclusion criteria were included in study. The participants comprised of 77 males and 34 females. The ratio of males to females was 2.26:1. The mean age of study participants was 28.42 ± 10.29 years and the mean height was 167.24 ± 10.63 cm.

The results of sway velocity measures collected from the two machines are shown in Table 1. It was seen that there is a serial increase in the sway velocity from condition I to IV on both the devices, indicative of the increasing level of difficulty from condition I to IV. One sample T-test was used to check if the mean differences in the values measured from the two devices were different than zero. It was seen that the results obtained from the two devices were significantly different (p < 0.005). Pearson’s correlation coefficient was used to compare the values measured by the two devices. The measured values from both platforms show a correlation coefficient ranging from 0.53 to 0.91 in the four test conditions, with very high—high correlation between the data pairs in test conditions I & II, and moderate correlation in test conditions III & IV. The best correlation (r = 0.91, R² = 0.83) was observed in test condition I. Pooling in the results of all four test conditions and comparing the pairs of values of BBM and WBB using linear regression shows an overall correlation coefficient of 0.62, indicative of a moderate positive correlation (Fig. 3a).

Table 1.

Results of sway velocity measured from the WBB versus BBM

Test condition	BBM (cm/s)	WBB (cm/s)	Mean difference (95% CI)	r (95% CI)
I	5.36 ± 0.93	2.17 ± 0.36	− 3.20 (− 3.31, − 3.08)	0.91 (0.88, 0.94)
II	5.38 ± 0.93	2.26 ± 0.36	− 3.12 (− 3.24, − 2.99)	0.79 (0.70, 0.85)
III	5.43 ± 0.87	2.46 ± 0.42	− 2.98 (− 3.10, − 2.85)	0.69 (0.57, 0.77)
IV	5.96 ± 0.95	3.52 ± 0.70	− 2.44 (− 2.60, − 2.28)	0.53 (0.38, 0.65)

Fig. 3.

a Linear regression of sway velocity from WBB versus BBM, b Bland Altmann plot of differences versus mean of sway velocities

Correlation was also studied between the other sway parameters calculated from the WBB and BBM (Table 2). It was seen that the computed values of sway area and 95% confidence ellipse area show a high positive correlation between the WBB and BBM. Overall, the sway area correlated best between the two devices (r = 0.75, R² = 0.56), while the 95% confidence ellipse area had a Pearson’s correlation coefficient of 0.73 and R² of 0.52, when comparing the values across all four test conditions.

Table 2.

Comparison of sway parameters measured from WBB versus BBM

	Sway velocity (cm/s)		r	Sway area (cm2/s)		r	CEA (cm2)		r
Test condition	BBM	WBB		BBM	WBB		BBM	WBB
I	5.36 ± 0.93	2.17 ± 0.36	0.91	0.59 ± 0.35	0.26 ± 0.12	0.52	1.46 ± 2.15	1.27 ± 1.10	0.49
II	5.38 ± 0.93	2.26 ± 0.36	0.79	0.52 ± 0.30	0.29 ± 0.16	0.69	0.91 ± 1.46	1.25 ± 1.46	0.61
III	5.43 ± 0.87	2.46 ± 0.42	0.69	1.13 ± 0.40	0.52 ± 0.26	0.47	4.67 ± 2.73	4.70 ± 4.82	0.52
IV	5.96 ± 0.95	3.52 ± 0.70	0.53	1.62 ± 0.69	1.28 ± 0.61	0.56	7.91 ± 5.21	14.27 ± 9.28	0.54
All readings	5.53 ± 0.95	2.60 ± 0.72	0.62	0.92 ± 0.66	0.58 ± 0.53	0.75	3.55 ± 4.23	5.33 ± 7.48	0.73

Bland Altmann scatter plot was used to compare the differences in sway velocities to the mean of the sway velocities determined from the two devices (Fig. 3b). The plots show that the WBB consistently underestimated the sway velocity as compared to the BBM. Most of the measurements were within 1.96*SD of the mean difference in measurements between the WBB and BBM, with a few outliers, implying a good agreement between the two devices. A comparison of all the 444 pairs of readings obtained by the two devices in all four test conditions shows that the WBB underestimates the sway by 2.93 cm/sec. A trend was also observed that the difference between the measurements by the two devices increases with increase in magnitude of sway velocity recorded by them.

The number of sway velocity values exceeding the reference normal values for the BBM were counted from the test reports generated by the NeuroCom Posturography software. The reference Mean ± SD of equilibrium scores provided by a previously published study were used to estimate the 95% Confidence Intervals (CI) of normal equilibrium score for WBB by the formula 95%CI = Mean ± 1.96*SD/√N [13]. The normal values are as per Table 3. The number of values of equilibrium score less than the estimated normal values for WBB were counted. A 2 × 2 contingency table was generated (Table 4) by counting the number of abnormal values of Posturography for the two devices. The sensitivity of Wii Balance Board in detecting a balance dysfunction was determined to be 69.39%, the specificity 73.16%, predictive value of positive test 24.29%, and that of negative test 95.07%, and an accuracy of 72.75%.

Table 3.

Known normal values of postural sway parameters for BBM and WBB

Test condition	BBM* (degrees/sec)			WBB# (equilibrium score)
	< 39 yrs	40–59 yrs	60–69 yrs	18–68 yrs
I	0.35	0.45	0.45	0.68
II	0.45	0.55	0.45	0.68
III	0.75	0.95	0.95	0.69
IV	1.95	2.35	2.45	0.71

*Values greater than reference sway velocity for BBM are considered abnormal

^#Values lesser than reference equilibrium score for WBB are considered abnormal

Table 4.

2 × 2 contingency table for abnormal postural sway detected from WBB versus BBM

	BBM
	Positive	Negative
WBB
Positive	34	106
Negative	15	289

Discussion

The importance of Posturography in evaluation of balance lies in its ability to provide a functional, objective and quantitative measure of stability of an individual. It has the capability to test the balance while separately eliminating sensory inputs of vision and proprioception, and also can be used to provide conflicting sensory inputs to further evaluate the postural stability in challenging environments.

The WBB was introduced into the market in 2007, ushering a new era of motion-controlled gaming, wherein the position of an individual standing on the gaming platform could be used to control motion in a video game. The technology behind it is similar to that used in static Posturography machines [5]. It was not long before the same caught the researcher’s attention and the WBB became the tool of research on balance by the neurologists and neuro-otologists.

However, the WBB is intended by the manufacturer as a gaming accessory, and is optimized for the same. It offers a cheap alternative to medical grade Posturography machines, but lacks the precision and calibrations needed for clinical Posturography. The limitations that have been highlighted in literature include inconsistent sampling rates, lack of precision, low signal to noise ratio and inability in detecting certain movements and shear forces [5]. The WBB is provided with Wii Fit software, but the scores derived from it have been found to lack validity [15, 16]. The researchers have therefore had to resort to custom programs to collect the data of centre of pressure from the WBB for Posturography computations [4, 10]. Several open source and free software have been developed by researchers from various institutions that can be used to collect and process the data from WBB [6, 8]. The same data can then be subjected to further mathematical calculations by software such as Scilab or Matlab and various measures of postural sway such as path length, sway velocity, sway area and 95% confidence ellipse area can be calculated [6, 12]. The BBM in its reports by default, only provides the sway velocity in degrees/s without calculating the other sway parameters. Therefore, further analysis and computation of sway parameters required exporting of raw Posturography data from both the platforms and using self-written program in Scilab (a free and open-source computation software) incorporating the mathematical formulae described earlier.

The results of our study show a moderate to high positive correlation between the measured sway velocities from WBB and BBM, ranging from 0.53 to 0.91 in various test conditions, with overall value of 0.62 for all pairs of values considered together. The results of our study are similar to those of Chang WD et al. who compared the path length measured by WBB to NeuroCom Smart Balance Master, and found a correlation coefficient ranging from 0.58 to 0.86 in different test conditions [17]. The overall moderate correlation is likely due to the lack of standardisation of the gaming platform, the lack of calibration, and lack of accuracy. As demonstrated by Leach JM et al. calibration of the WBB results in improved accuracy of its results [10]. Thus, the results of postural sway as calculated from the gaming platform are likely to vary from those obtained from medical grade Posturography machines.

There have been several studies investigating the reliability and validity of the Wii Balance Board as a Posturography tool. Rey-Martinez et al. studied the validity of results obtained from the Wii Balance Board versus a standard Posturography machine (Balance Rehabilitation Unit). They found an overall Intra-class correlation coefficient (ICC) of 0.94 for all the test conditions. They also found that higher ICC values were obtained in complex conditions, when the study subjects were standing on foam placed over the force plate than in simple conditions, when the subjects were standing directly on the force plate [6]. The results of the current study however show an inverse trend, with the highest values obtained in test conditions I and II, as compared to the lower values in test conditions III and IV, possibly due to the difference in medical grade Posturography machine used in the different studies.

The Bland Altmann plot (Fig. 3b) shows that the WBB consistently underestimates the sway as compared to the BBM. Previous studies using the Bertec force plate as standard instead of the BBM have shown that the WBB overestimates the postural sway as compared to laboratory grade force plates [18, 19]. While a study utilizing the Zebris force plate as standard found that WBB underestimates the path length as compared to the Zebris force plate [20]. The differences in estimation of sway velocity by WBB as compared to the standard are thus likely due to the difference in laboratory grade force plate used by the studies. It is also seen that the difference in sway velocity measured by the two instruments increases with the increase of the magnitude of measured values. The results of our study are similar to those obtained by Severini G et al. who found that the accuracy of WBB with respect to medical grade force plate decreases with increase in value of estimated sway [19].

Despite a large number of studies investigating the reliability and validity postural sway measured from WBB, none have investigated its ability in detecting abnormal sway as compared to medical grade Posturography machines [5]. Literature search found only one study which has defined the normative values for Posturography results obtained from WBB [13]. The results of our study show a 69.39% sensitivity in detecting abnormal postural sway, a 73.16% specificity in ruling out abnormal postural sway along with 24.29% and 95.07% positive and negative predictive values for WBB respectively. Thus, the postural sway measured from the WBB if found to be within the normal range can be regarded as normal while sway beyond the normal limits would require further verification by other clinical tests. This aspect of using the Wii Balance Board for detecting abnormal postural sway has not been stated in prior studies and will require further research for validation.

The current study has a number of limitations. The tests were done on apparently healthy adults without any known balance dysfunction. The results require further validation for persons with known balance issues and neurological illnesses. The tests on two instruments were done successively rather than simultaneously, which can account for variability between the two measures. The study protocol only used mCTSIB as the balance test. Other tests of Posturography such as the limits of stability, unilateral stance and rhythmic weight shift were not used for simplicity of study design. The normative values for WBB in the current study were estimated from previously known Mean ± SD values of equilibrium scores stated by Domènech-Vadillo E et al. Normative values for WBB in terms of other reported parameters of postural sway will need to be established by further studies collecting Posturography data of normal adults.

From the results of our study, we can conclude that the Nintendo Wii Balance Board can be used as a low-cost and satisfactory solution for objective analysis of balance when high precision of measurement is not required. Owing to the moderate sensitivity and specificity, low positive and high negative predictive values, it is likely to be useful for ruling out balance dysfunction rather than diagnosing balance dysfunction.

Declarations

Conflicts of interest

The authors have none to declare.

Ethical Approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed Consent

Informed consent was obtained from all individual participants included in the study.