Effect of Head-Mounted Display Virtual Reality with Visual Intervention on Motion Sickness Symptoms

2.1. Head-mounted Display Virtual Reality (HMD VR) / Ekran virtualne stvarnosti nataknut na glavu

VR is a technology based on computer graphics, which allows the creation of virtual scenes and objects to be manipulated by users through input devices. This simulation can be experienced scenarios through output devices that engage senses such as sound, touch, and smell while users feel high reality during interaction. VR combines three main aspects namely immersion, interaction, and imagination as shown in Figure 1. Moreover, creating a sense of presence is a major motivation for using VR in education and other domains. This research examines several factors influencing users interaction and the result shows that deficiencies in visual presentation, such as poor graphics, reduce the sense of reality experienced in VR [13].

Figure 1 Dimensions of VR

Slika 1. Dimenzije virtualne stvarnosti

Previous research has shown that people using HMD VR with darker visual are more prone to experience MS than those with brighter graphics. Other influential factors includes the quality of graphics shown in VR as lower-quality or rough graphics tend to increase the probability of MS compared to smoother, higher-quality graphics [14].

2.2. Motion Sickness (MS) / Morska bolest

MS occurs when sensory information from the inner ear fails to synchronize correctly between different sensors. Symptoms often mistaken for fatigue or boredom are triggered by motion exposure. Importantly, without the presence of nausea and vomiting, a decrease in performance may go unrecognized as a sign of MS [15]. Experimental results have shown that nearly all people with normal vestibular function experience some level of MS when exposed to provocative physical body movements, vestibular-ocular reflex disorders, or optokinetic stimulation [16].

2.3. The Influence of Color on Human Psychology / Utjecaj boja na ljudsku psihu

Color psychology is based on mental and emotional effect that colors have on humans in various aspects of life. In color therapy, colors are often connected to emotion which can affect mental or physical state of a person [17]. The influences of colors are generally observed across two main categories, namely warm and cold tones. Warm tones are colors with red end of the spectrum are categorized as warm colors, including red, yellow, and orange. These colors can cause a wide range of emotional reactions from comfort, and warmth to anger as well as violence. On the other hand, cold tones are are considered more calming compared to warm color category. The colors in the cold tone category include green, blue, and purple. In addition, the influence of color on human psychology often causes feelings of calm as well as sadness.

3.1. Sailing Route and Duration / Plovidbena ruta i trajanje

Passenger ships are designed to transport people from one location to another. During the design of a ship, stability would be crucial for safety, as it ensured balanced between gravitational force and buoyancy. Moreover, the movement of the ship could lead to MS among passengers. In this research, test was conducted on ship for passengers sailing from Tanjung Perak Port Surabaya, Indonesia to Gili Mas Port Lombok, Indonesia) with a total duration of 20 hours. Sea conditions on this route were relatively calm with an average wave height ranging from minor (0-0.1 m) to medium (1.25-2.5 m). The VR experience for passengers boarding on this route was designed to last for 14 minutes. This duration was selected based on research by Santoso [18], stating that 10% of passengers showed the first MS symptoms after 30 minutes, 10% after 2 hours, and 10% after 8 hours, respectively. Therefore, the duration of sailing effectively covered all these time frames. The overall route of the ship at the time of the experiment was shown in Figure 2.

3.2. Experiment Variables / Varijable eksperimenta

This research cumulatively showed the process and results conducted in different stages. In stage 1, HMD VR was used with a causative approach to evaluate the effect of blurriness on the severity of MS. Following that, a symptomatic method was used in the second stage of this research to examine MS in passenger ships with visual aspect of blurriness. This research marked as the third stage, added the color tone and passengers’ position as independent variables.

3.2.1. Free variables / Slobodne varijable

MS was closely related to the physical movements experienced by human body. While passengers were onboard of a sailing vessel, an individual body moved in response to the movement of ship which was influenced by sea current, wave, or wind. During the voyage, the ship shows six degrees of freedom (DOF) including heave, sway, surge, roll, pitch, and yaw. Wertheim et al. [19] explored the effect of combinations of heaving, pitching, and rolling motion on individuals. The research aimed to test the traditional assumption that MS on a vessel was primarily triggered by the heave motion, while the pitch and roll motions were considered ineffective. However, this research did not consider motion of ship in the water as the experiment was performed in real environment rather than a simulator. Consequently, motion of ship during the investigation was considered as free variable.

Figure 2 Ship Route during the experiment

Slika 2. Ruta broda za vrijeme eksperimenta

3.2.2. Dependent variables / Ovisne varijable

Table 1 showed the review of all sample conditions based on dependent variables before and after the experiment. The description of each dependent variable is explained as follows.

According to Lawther & Griffin [20], children aged 6 to 7 years started to experience MS, with symptoms peaking around ages 9 to 10 years. These symptoms began to disappear when the children reached 15 years old, and the tendency to experience MS decreased into adolescence and adulthood. Given that most passengers on the ship were between 24 and 40 years old and had traveled by ship at least once, this research focused on passengers aged 15 to 40 years. This age range was selected because the selection represented a group with lower tendency for MS and was commonly found among ship passengers.

N. Sugita et al. [21] investigated effect of visually-induced motion sickness (VIMS) by analyzing causal coherence functions between blood pressure and other variables. This research found significant differences in both traditional and causal coherence functions between individuals who experienced MS and healthy people. The relationship between cardiovascular system indices, such as heart rate and blood pressure, changes due to alteration in autonomic nervous activity caused by visual stimulation. The coherence function between blood pressure and heart rate along with two causal coherence functions showed the total linearity of the cardiovascular system. This process showed that blood pressure was an important dependent variable in assessing MS. Following this discussion, blood pressure should be measured before and after using HMD VR.

Susceptibility to MS was influenced by the consumption of alcohol and the use of relief tablets. Previous research showed that alcohol, as signified by Blood Alcohol Concentration (BAC) had a negative relationship with the severity of MS. Additionally, [12] stated that participants in the experiment should avoid taking MS relief tablets, as the tablets were used to prevent MS and thereby reduced the severity of symptoms during travel.

Table 1 Review of sample condition before and after the experiment

Tablica 1. Pregled stanja uzorka prije i nakon eksperimenta

3.2.3. Independent Variables / Neovisne varijable

VR model incorporated three variables including image quality (clear and blurry), visual intervention with color tones, and passengers’ positions as shown in Figure 3. Following the discussion, these variables were tested on ship passengers participating in this research. The data collected via Sickness Simulator Questionnaire (SSQ) was then tested using three methods, namely ANOVA Multiway, Multiple Linear Regression, and paired t-test.

Figure 3 VR Model

Slika 3. Model virtualne stvarnosti

The samples for this research were collected from passengers onboarding a passenger ship sailing from Tanjung Perak Port to Gili Mas Port in Lombok. All criteria mentioned were then interpreted into standards that were met by all participants in the following manner.

All participants should be willing to participate in experiments in various conditions. Due to the difficulty during finding participants that met these requirements, the experiment was designed to include three samples in each group, following Least Significant Difference (LSD) Formula [22].

3.3. Sickness Simulator Questionnaire SSQ / Upitnik o simulatoru bolesti

The research used SSQ because it effectively captured subjective experiences of MS without requiring participants to vomit. This method was useful when handling small sample sizes such as passengers prone to vomiting. MS symptoms often resembled typical MS, but effects of simulator sickness were generally milder and less severe among exposed populations. SSQ templates developed by Kennedy, et al. classified tissue disorders across several dimensions, including nausea, oculomotor, and disorientation [23].

Table 2 Independent variables of the experiment

Tablica 2. Neovisne varijable eksperimenta

Table 3 showed how to calculate MS symptoms results for Nausea (N), Oculomotor (O), Disorientation (D), and the resulting Total Score. The results of the reduction between symptoms before and after VR was used produced the final score on SSQ.

Table 3 Sickness Simulator Questionnaire (SSQ)

Tablica 3. Upitnik o simulatoru bolesti

Table 4 Equations for calculating the SSQ Component Value

Tablica 4. Jednadžbe za izračun SSQ vrijednost komponente

3.4. Experimental Data / Eksperimentalni podaci

Table 5 showed the experiment procedure used to obtain data that was conducted on passengers of the ship.

Table 5 Experiment Activity and Duration

Tablica 5. Aktivnost eksperimenta i trajanje

3.5. Statistical Data Analysis / Statistički podaci i analiza

Several statistical tests were conducted to determine the relationship between each variable and symptoms of MS experienced by passengers using HMD VR. Although the main statistical test was three-way ANOVA, other methods such as Wilcoxon test and multiple linear regression were also used to validate the result of ANOVA.

3.5.1. Analysis of Variance (ANOVA) / Analiza varijacija

Three-way ANOVA was used due to the presence of three independent variables. This method was used to determine whether the various criteria tested influenced the desired results.

Performing three-way ANOVA required preliminary tests to validate the sample data. This research used three classic assumption tests, namely normality test, homogeneity test, and independence test. The samples were independent and randomly selected from each population or data group. When the assumption test was satisfied, ANOVA was then performed. Moreover, the results of ANOVA analysis showed whether there was a significant change in symptoms of MS before and after using VR. The classic assumption tests that were met included the following,

The examination was conducted to determine when the data obtained was normally distributed. The basis for decision-making was,

The examination aimed to explore whether the sample group had the same variance. The basis for decision-making for this test included the following.

This examination aimed to determine the correlation among the experimental data. Additionally, experimental data should not have a significant correlation with other data to be independent. In this ANOVA test, the following calculation formula was used.

[TeX:] JKT = \sum_{i = 1}^{r}{\sum_{j = 1}^{c}{x_{\text{ij}}^{2} - \frac{T_{\text{..}}^{2}}{\text{rc}}}}$JKT={\sum }_{i=1}^r{\sum }_{j=1}^c{x}_{\mathrm{\text{ij}}}^2-\frac{T_{\mathrm{\text{..}}}^2}{\mathrm{\text{rc}}}$ (1)

[TeX:] JKB = \frac{\sum_{i = 1}^{r}T_{i}^{2}}{c} - \frac{T_{\text{..}}^{2}}{\text{rc}}$JKB=\frac{{\sum }_{i=1}^r{T}_i^2}{c}-\frac{T_{\mathrm{\text{..}}}^2}{\mathrm{\text{rc}}}$ (2)

[TeX:] JKK = \frac{\sum_{j = 1}^{c}j^{2}}{r} - \frac{T_{\text{..}}^{2}}{\text{rc}}$JKK=\frac{{\sum }_{j=1}^c{j}^2}{r}-\frac{T_{\mathrm{\text{..}}}^2}{\mathrm{\text{rc}}}$ (3)

[TeX:] JKG = JKT - JKB - JKK$JKG=JKT-JKB-JKK$ (4)

where: JKT : Total Squared Sum JKB : sum of squares of rows JKK : sum of squares of columns JKG : sum of squares error

3.5.2. Wilcoxon / Wilcoxon

This analysis was part of a nonparametric test, which did not require normally distributed data. In addition, the test was conducted as a substitute for ANOVA analysis. Since the assumption of the parametric ANOVA test was not met, the nonparametric test was performed. The results of this test determined whether there was a significant difference between the independent variables on SSQ. Relating to the analysis, the formula of Wilcoxon test was as follows.

[TeX:] Z = \ \frac{J - \ \mu J}{\text{σJ}} = \ \frac{J - \ \frac{n(n + 1)}{4}}{\sqrt{\frac{n(n + 1)(2n + 1)}{24}}}$Z=\frac{J-\mu J}{\mathrm{\text{σJ}}}=\frac{J-\frac{n(n+1)}{4}}{\sqrt{\frac{n(n+1)(2n+1)}{24}}}$ (5)

where: Z : Wilcoxon z-test J : Small number of levels μJ : Average value σJ : Standard deviation value n : number of samples

The basis for decision-making for Wilcoxon test was as follows.

3.5.3. Multiple Linear Regression / Višestruka linearna regresija

Multiple linear regression was used when the number of independent variables or predictor variables was more than one. During this process, the analysis produced values from both F-test and t-test. F-test was used to analyze the collective effect of independent variables on dependent variable, while t-test was used to determine the partial effect of each independent variable on dependent variable. The multiple linear regression model was represented by the following equation.

Y = α + β₁ X₂ + β₂ X₂ + β_n X_n + e (6)

where: Y : Dependent variable or response variable X : Independent variable or predictor variable α : Constant β : Slope or Coefficient estimate

The basic assumption tests that were met in Multiple Linear Regression analysis included the following.

This analysis showed a perfect linear relationship between some or all the variables that explained regression model. The basis for decision-making for the test included the following.

Based on tolerance value

Based on the variance inflation factor (VIF) value

This test was part of the classical assumption test aimed to determine the correlation of confounding errors in period t with previous period t-1 in linear regression model. When correlation was detected in the data, it was referred to as an autocorrelation problem in the analysis data. The basis for decision-making in the analysis included the following statements.

Normality test was performed to determine the data distribution in this research. When the data was abnormally distributed, linear regression method could not be applied because the prerequisites of the classical assumption test were not met. Following this process, two methods were used for this test, namely Kolmogorov-Smirnov test for samples more than 50 and Shapiro-Wilk test for samples less than 50. Since the data in this research was less than 30, normality test was conducted using the Shapiro-Wilk test.

Homoscedasticity test aimed to determine any deviation from the classical assumption of heteroscedasticity, which referred to the presence of variance inequality in residuals across all observations in regression model. Glejser test method was used in this research as the method provided more accurate results compared to plot test which could cause bias. The examination was conducted by regressing independent variable on its absolute residual value of dependent variable.

3.5.4. Paired T-test / Upareni test

Paired t-test was used to compare two samples from the same subject in different treatments. This model analyzed differences before and after treatment, allowing the model to be a common method to assess the effectiveness of treatment. The efficiency was shown by the changes in the average values before and after the treatment [24]. Moreover, Paired t-test included parametric tests, showing that assumptions should be met.

The paired t-test in this investigation had the following hypothesis.

The test statistics in this research included the following equation.

[TeX:] t_{\text{hit}} = \frac{\overline{D}}{\frac{\text{SD}}{\sqrt{n}}}$t_{\mathrm{\text{hit}}}=\frac{\overline{D}}{\frac{\mathrm{\text{SD}}}{\sqrt{n}}}$ (7)

[TeX:] SD = \sqrt{\text{var}}$SD=\sqrt{\mathrm{\text{var}}}$ (8)

[TeX:] var = \frac{1}{n - 1}\sum_{i = 1}^{n}{(x_{i} - \overline{x})}^{2}$var=\frac{1}{n-1}{\sum }_{i=1}^n{(x_i-\overline{x})}^2$(9)

where,

[TeX:] t_{\text{hit}}$t_{\mathrm{\text{hit}}}$ = Calculated value,

[TeX:] \overline{D}$\overline{D}$ = average difference between measurements 1 and 2

[TeX:] \text{SD}$\mathrm{\text{SD}}$ = standard deviation difference between measurements 1 and 2

[TeX:] \text{var\ \ }$\mathrm{\text{var}}$ = variance ([TeX:] s^{2}$s^2$)

[TeX:] n$n$ = number of samples

4.1. Classical Assumption Test / Klasični test procjene

Classical assumption test was a prerequisite in statistical analysis to ensure the validity of the model. This examination was essential in ANOVA and Multiple Linear Regression analysis, where assumption test was conducted before proceeding with the analysis. The results of the experiments were analyzed to determine whether the data met the criteria. The explanation of some classical assumption tests was as follows.

Normality Test was used to determine whether the data was normally distributed. This assumption was evaluated using two methods, namely Kolmogorov-Smirnov test for sample size greater than 50 and Shapiro Wilk test for data samples less than 50.

The results of residual normality and normality tests using Shapiro Wilk method in SPSS showed (significance value) Sig. = 0.687 and 0.127. Since Sig. > 0.05, the data in this experiment was considered normally distributed.

Table 6 Normality Test Comparison

Tablica 6. Usporedba testa normalnosti

Homogeneity test aimed to determine whether the variation of some data from the population had the same variance. The assumption was accepted when the variance of several populations was equal or homogeneous. The basis for decision-making in this test was based on Sig. at mean > 0.05, then the data was considered homogeneous.

Table 7 Homogeneity Test Results on Color Type Variables

Tablica 7. Rezultati testa homogenosti na varijable vrsta boje

When Sig. < 0.05, the result showed that the variance of data from all variables was not equal or homogeneous. Although Homogeneity assumption test of data variance was not met, ANOVA test was still conducted, because the number of samples in this experiment was balanced. Therefore, conducting independence test was performed during the investigation.

Independence test in this research was conducted as a parametric test for independent comparison. Following this discussion, an independent sample in the results produced data from different subjects.

Table 8 Independence Test Results

Tablica 8. Neovisni rezultati testa

Sig. (2-tailed) value of image quality variable > 0.05 and the color type as well as position variables of passengers < 0.05. The values showed that there was no difference between SSQ score and image quality variable, but there was a difference between variable color type and position of passengers.

Table 9 Descriptive Statistics on Independent Variables

Tablica 9. Opisna statistika neovisnih varijabli

The lower average value on independent variable shown in the Table implied that the factors had a better potential to reduce MS symptoms.

Since some assumption tests such as homogeneity test and independence test were not satisfied, ANOVA analysis was not proceeded. The analysis was performed using a nonparametric Wilcoxon test, which did not rely on assumption tests and was used to compare more general groups.

Multicollinearity test was conducted to determine whether there was a correlation between independent variables in regression model.

Table 10 Multicollinearity Test Results

Tablica 10. Rezultati testa višebojne linearnosti

The test results showed the value of tolerance > 0.10 and value of VIF < 10.00, implying that non-multicollinearity was met. This process implied the data did not experience multicollinearity problems.

This test was represented using Durbin Watson values which were one method for evaluating autocorrelation. According to the decision-making criteria, non-autocorrelation was satisfied when dL < d < 4-dU.

Table 11 Durbin Watson Autocorrelation Test Results

Tablica 11. Rezultati Durbin Watsonova autokorelacijskog testa

To determine dU and dL values, Durbin Watson Tables were used. For sample size N of twenty-four, and three number of independent variables (k), dL and dU values were 1.101 and 1.656. In this research, Durbin Watson value obtained was 2.272, hence, the data did not experience autocorrelation problems.

The working principle of Glejser test was to progress independent variables to absolute residual values or ABS_RES. In this research, homoscedasticity was satisfied when Sig. > 0,05.

Table 12 Homoscedasticity Test Results

Tablica 12. Rezultati testa homogenosti varijacija

At the value of Sig., image quality variable = 0.954, color type variable = 0.333, and position variable = 0.219. All variables showed homogeneous symptoms with Sig. > 0.05, implying the data satisfied the assumption of homoscedasticity.

4.2. Results / Rezultati

After performing all assumption tests and results showed that all conditions were passed, parametric statistical tests of Multiple Linear Regression, Paired t-test, and Wilcoxon nonparametric statistical tests were conducted.

Table 13 F-Test Results

Tablica 13. Rezultati testa

The Table showed that at Sig., 0.000 < 0.05 in F-test trigger. Therefore, independent variable simultaneously contributed to the influence on dependent variable of MSsymp toms.

Table 14 T-test Results

Tablica 14. Rezultati T-testa

In the results of the t-test, image quality value = 0.104 greater than 0.05, signifying that image quality did not have a significant effect on SSQ score. Consequently, the values of color type and position are both 0.000, which was less than 0.05. This result showed both color type and passengers position had a significant effect on SSQ score as MS symptoms was measured.

Table 15 Coefficient of Determination Test Result

Tablica 15. Koeficijent određivanja rezultata testa

The results of coefficient of determination test were represented by R Square value, which was 83.7%. This result implied the independent variable simultaneously contributed to the influence of dependent variable of MS symptoms by 83.7%. The remaining 16.3% was influenced by other variables not included in this equation or variables not studied.

Table 16 Paired T-test Analysis Results

Tablica 16. Upareni rezultati T-testa

Sig. (2-tailed) values were 0.021 < 0.05 in paired t-test analysis. This result was interpreted that H₀ was rejected and H₁ was accepted. Consequently, there was a significant difference between before and after getting intervention for the treatment.

Table 17 Wilcoxon Analysis Results

Tablica 17. Rezultati Wilcoxon analize

During analysis of Wilcoxon, which was a nonparametric statistical method, Asymp. Sig. (2-tailed) value was 0.028 < 0.05. Therefore, the result showed that H₀ was accepted signifying a difference between before and after getting intervention for the treatment.