Clinical predictors of cybersickness in virtual reality (VR) among highly stressed people

The use of virtual reality (VR) in the treatment of psychiatric disorders is increasing, and cybersickness has emerged as an important obstacle to overcome. However, the clinical factors affecting cybersickness are still not well understood. In this study, we investigated clinical predictors and adaptation effect of cybersickness during VR application in highly stressed people. Eighty-three healthy adult participants with high stress level were recruited. At baseline, we conducted psychiatric, ophthalmologic, and otologic evaluations and extracted physiological parameters. We divided the participants into two groups according to the order of exposure to VR videos with different degrees of shaking and repetitively administered the Simulator Sickness Questionnaire (SSQ) and the Fast Motion sickness Scale (FMS). There was no significant difference in changes in the SSQ or the FMS between groups. The 40–59 years age group showed a greater increase in FMS compared to the 19–39 years age group. Smoking was negatively associated with cybersickness, and a high Positive Affect and Negative Affect Schedule score was positively associated with cybersickness. In conclusion, changing the intensity of shaking in VR did not affect cybersickness. While smoking was a protective factor, more expression of affect was a risk factor for cybersickness.


Methods
Participants. We recruited 83 healthy adult volunteers at least 19 years of age with high stress level from October 2016 to January 2018. We defined high stress as a score of 20 or more on the Perceived Stress Scale-10 (PSS-10) 29 . Given that the occurrence of cybersickness is affected by individual factors and VR has been often used and researched for the reduction of anxiety or stress in psychiatry, we selected the target population as highly stressed but healthy adults in this study. Inclusion criteria were healthy persons who voluntarily participated in this study and had no problems understanding study procedures and controlling VR equipment. Those who had major depressive disorder, bipolar disorder, schizophrenia, other psychotic disorders, delusional disorder, anxiety disorders, delirium, dementia, eating disorders, alcohol use disorder, organic mental disorders, mental retardation, psychiatric disorders due to medical conditions, or suicidal risk were excluded from the study. In addition, those who had neurological illnesses such as stroke or epilepsy and serious medical illnesses were excluded. Those who had medical or surgical history of psychiatric, otologic, or ophthalmologic disorders or problems with neck movements were also excluded. All participants were drug-naive when sample measurement was conducted at the baseline evaluation. At the baseline screening visit, participants were evaluated independently by the psychiatrist (HJ Jeon), the otorhinolaryngologist (WH Chung), and the ophthalmologist (K Park). In order to evaluate psychiatric disorders, a psychologist who specialized in psychiatric evaluation administered the Korean version of the Mini International Neuropsychiatric Interview (MINI) 30 to the subjects, according to the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) 31 . The study was approved by the Institutional Review Board of the Samsung Medical Center, all experiments were performed in accordance with Good Clinical Practice guidelines, and all participants gave written informed consent at enrollment into the study.
Physiological parameters. Using a computerized biofeedback system, ProComp Infiniti 37 (Thought Technology Ltd., Montreal, Canada), physiological parameters were obtained through sensors attached to each subject's body. These physiological parameters included electromyography (EMG), skin conductance, skin temperature, respiration amplitude, and heart rate/blood vessel pressure (HR/BVP), along with heart rate variability (HRV) parameters such as the HR from the inter-beat interval (IBI), very low frequency band (VLF), low frequency band (LF), high frequency band (HF), LF/HF ratio, number of interval differences of successive normal-tonormal (NN) intervals greater than 50 ms (NN50), standard deviation of NN (SDNN), and the root mean square of the successive differences (RMSSD). Parameters extracted through monitoring for 3 min and 30 s at baseline were included in the analysis. EMG was recorded through the surface EMG sensors those were placed on the skin's surface. The sensors can record EMG signals of up to 1600 microvolts (µV), root mean square and measure muscle activity. Skin conductance was measured through two electrodes those were strapped to two fingers of one hand. Skin conductance represents changes in the sympathetic nervous system. When a subject becomes more or less stressed, the skin conductance increases or decreases proportionally. Skin temperature was recorded through thermistor, which were strapped to the dorsal or palmar side of a finger. The peripheral temperature varies according to the amount of blood perfusing the skin and dependent on a subject's state of sympathetic arousal. As a subject gets stressed, their extremities tend to get colder. Respiration amplitude was recorded through the respiration sensor that were strapped around a subject's abdomen. It detects the expansion and Application of VR. Participants were exposed to shaking and dizzy immersive VR videos while sitting on a chair. The original video was provided by Korea Land and Geospatial Informatix Corporation and artificially modified for this study by adding a roll swing of the sine waveform at 30 Hz in the Z-axis direction with 0.008°/s for each grade (Fig. 1). Then we made image movements of 0.3°/s (VR with less shaking) and 0.38°/s (VR with more shaking). Participants were exposed to a VR video that involved walking on a shaky path for 3 min and 30 s, and after a break of 3 min and 30 s, they were exposed to another VR video that differed in the intensity of shaking from the first video. Among the total study group, VR was applied in the order of escalating degrees of shaking for 40 people and in the order of de-escalating degrees of shaking for the other 43 people (Supplementary Fig. 1). During the exposure to the VR videos, participants were asked to count the number of persons who appeared in the video in order to increase their attention to it. The study was conducted in a room that was exclusively prepared to block outside noise in the Clinical Trial Center located in Samsung Medical Center. Samsung Gear VR (Samsung Electronics Co., Ltd., Suwon, South Korea) was used for the study, and the head-mounted display (HMD) device included separate screens for each eye, integrated head-tracking, and stereo earphones.

and the Fast Motion
Sickness Scale (FMS) 39 were used. The SSQ and FMS were administered before VR application and again immediately after application of each VR video. The primary outcomes were the changes in SSQ and FMS scores.

Statistical analyses.
We examined the distribution of demographic characteristics, baseline psychological scales, physiological parameters, ophthalmologic parameters, and otologic parameters. Categorical variables were presented as frequencies with percentages and continuous variables as means with standard deviations (SD). The changes in the SSQ and FMS according to the order of exposure to VR videos, age, and sex were analyzed using either a Student's t-test or an analysis of variance (ANOVA). To confirm the correlation between potential factors and the changes in SSQ or FMS, Pearson's correlation or Spearman's rank correlation was used depending on the characteristics of variables. Variables with a p-value below 0.10 were included in multivariable regression analyses. We reported β-coefficients and p-values. We considered a p-value of less than 0.05 as statistically significant. All statistical analyses were performed using IBM SPSS Statistics Software (version 24; IBM, New York, USA).

Results
Baseline characteristics of participants.  Table 2 shows the change in SSQ and FMS after application of VR according to the order of exposure to VR, age group, and sex. There was no significant difference in the changes in SSQ and FMS between the group exposed to VR videos in order of increasing degrees of shaking and the group exposed to VR videos in order of decreasing degrees of shaking. According to age group, although there was no difference in the change of SSQ between groups, the changes in FMS in the 19-29, 30-39, 40-49, and 50-59 age groups were 3.87 (SD = 3.42), 4.68 (SD = 4.80), 7.84 (SD = 5.71), and 7.65 (SD = 3.77), respectively, and there was a significant difference between groups with p-value of 0.009. When age groups were divided as 19-39 and 40-59, the difference between age groups was more evident, with the changes in FMS being 4.28 (SD = 4.14) and 7.76 (SD = 4.09), respectively (p-value = 0.001).
According to sex, the SSQ increased by 29.60 (SD = 49.99) in males and 36.57 (SD = 40.69) in females. The FMS increased by 4.93 (SD = 4.23) in males and 6.98 (SD = 5.18) in females. Both SSQ and FMS increased more in females, but the differences with males were not statistically significant.

Correlation between factors and cybersickness.
We confirmed the correlation between variables and the change of SSQ or FMS. Among all clinical and physiological variables, smoking, NRS, PANAS-total, PANASpositive affect, PANAS-negative affect, NPA, and NPC showed significant correlations with the change in SSQ, while smoking, age, PANAS-total, and NPC showed significant correlations with the change in FMS (Table 3).

Discussion
In this study, we identified the adaptation effect according to varying degrees of shaking of VR videos and clinical predictive factors of cybersickness. When participants were exposed to two VR videos with different degrees of shaking in different orders, there was no difference in cybersickness between groups. There was a difference in the occurrence of cybersickness according to age group, and it was higher in the 40-59 age group compared to the 19-39 age group. Multivariate regression showed smoking was a factor that prevented cybersickness, and a high PANAS score was identified as a risk factor for cybersickness.
In this study, changing the intensity of shaking of VR videos did not affect cybersickness. Although previous studies have demonstrated adaptation effect according to exposure time 27 or repetition of exposure to VR 28 , this study did not show adaptation effect when the intensity of shaking of VR was increased or decreased. These findings suggest that regarding adaptation or habituation of cybersickness, exposure time to VR affect more sensitively than changing in intensity of sensory stimuli. The effect of the interaction between the intensity of stimulation and exposure time on cybersickness can be investigated in future studies. Table 1. Baseline characteristics of total participants. SD standard deviation, SSQ Simulator Sickness Questionnaire, STAI State-Trait Anxiety Inventory, NRS Numeric Rating Scale, PSS-10 Perceived Stress Scale, PANAS Positive and Negative Affect Schedule, SDS Sheehan Disability Scale, EQ-5D-5L Five-level version of EQ-5D, EMG electromyography, HR/BVP heart rate/blood vessel pressure, IBI inter-beat interval, VLF very low frequency band, LF low frequency band, HF high frequency band, HRV heart rate variability, NN50 number of interval differences of successive normal-to-normal (NN) intervals greater than 50 ms, SDNN standard deviation of NN, RMSSD the root mean square of the successive differences, TBUT tear breakup time, VHIT video head impulse test, SOT sensory organization test. In this study, we tried to determine clinical predictors of cybersickness through psychiatric, ophthalmologic, and otological evaluation and extraction of physiological parameters. Among ophthalmological and otological parameters, although NPA and NPC in pupils showed correlations with changes of the SSQ and the FMS, multivariate regression analyses did not show significance after adjusting the other factors. This finding suggests ophthalmological or otological impairment does not aggravate cybersickness. Previous studies have reported that motion sickness occur mostly in people with intact vestibular system 40,41 . Patients with impaired labyrinthine function do not normally experience classical motion sickness, but partially experience visually-induced motion sickness 40,41 . Likewise, recent evidence reports that individuals with a greater sensitivity to visual stimuli are more likely to experience more discomfort during VR applications 42 .
Previous studies have shown that smokers tend to have less motion sickness or postoperative nausea and vomiting, while nicotine nasal spray increases sensitivity to motion sickness. As an explanation of this, temporary nicotine withdrawal may lead to increased tolerance for motion sickness or nausea 43,44 . Nicotine affects motion sickness through the mechanism of the nicotinic cholinergic receptor (nAchR), which regulates the excitability of the caudal vestibular nucleus (CVN) 45 . The CVN contributes to both cardiovascular controls during head movements and autonomic manifestations of motion sickness through its strong connection with brain stem autonomic areas, such as the solitary tract nucleus and the parabrachial nucleus 46-50 . Smoking was a protective factor in our study, and it is thought that the short-term deprivation of nicotine may affect the result by this mechanism.
The high baseline score of PANAS was associated with an increased risk of cybersickness. All three PANAS variables, the total score, positive affect, and negative affect, showed significant correlations with SSQ, and in regression analyses, the total score of PANAS showed a significant association with changes in both SSQ and FMS. Also, the positive affect score of PANAS was associated with a change in SSQ in regression analysis. Because PANAS represents the expression of both positive and negative affect, our results suggest that more expression of affect is associated with cybersickness. Also, although there was no significant association with change in SSQ or FMS, the baseline score of 0-100 NRS, which measures subjective discomfort, was positively correlated with the change in SSQ. In previous studies, although the evidence of an association between cybersickness and affective expression or subjective discomfort is lacking, there has been evidence that emotional distress is associated with cybersickness. A study showed that there was no difference in simulator-related side effects between groups when exposure therapy was performed both conventionally and by VR in PTSD patients, and the authors suggested that it is possible that anxiety rather than VR accounts for any simulator-related side effects 51 . Another study found that anxiety has a mediating effect on cybersickness that occurs during VR application 52 . In addition, anxietyrelated personality traits are known to affect visual and vestibular control of balance [53][54][55][56][57][58] . Another study found that being in a VR does not cause anxiety by itself, but simulated motion can lead to anxiety 59 . There is also an evidence that the neuroticism personality trait is a mediating factor in the relationship between anxiety and the visuo-vestibular system. A study conducted that used a VR rollercoaster task found that neuroticism modulates the brain visuo-vestibular and anxiety systems during VR application 60 . In addition, in patients with persistent postural perceptual dizziness (PPPD), characterized by persistent dizziness and unsteadiness and exacerbated by upright posture, self-motion, and exposure to complex or moving visual stimuli, the neurotic personality trait was associated with brain regions mediating attention to visual motion cues 61 . These studies suggest that an individual's personality traits and anxiety may be more decisive predictors of cybersickness than the visuovestibular system. Given that an emotional state such as anxiety and personality traits such as neuroticism are related to cybersickness and influence each other, physicians should select VR content carefully, especially when using VR for the reduction of anxiety or stress.
This study has several limitations. First, since this study targeted a high stress group, it is difficult to apply the results of this study directly to the general population. However, if VR is used for anxiety or stress reduction, it is likely that users will have high stress. Moreover, considering that cybersickness is related to emotional expression or distress, results in the high stress group may be more applicable in clinical settings. In this study, screening was performed using the PSS-10, but in future studies, research on cybersickness in various target groups is expected to broaden accessibility to VR. Second, we investigated the adaptation effect using two VR videos with different degrees of shaking for 3 min and 30 s each, and there was no adaptation effect observed by this method. However, we should not conclude that there is no adaptation effect in cybersickness during VR application. According to previous studies, the presence of an adaptation effect depends on the methodology. In particular, when the same VR contents were applied, an adaptation effect appeared when the exposure time was prolonged. In the present study, the exposure time to each VR video was 3 min and 30 s, and it is likely that www.nature.com/scientificreports/ the specific methodology such as duration of exposure or degree of shaking will determine if there is an adaptation effect. Third, we applied VR to the subjects for a total of 7 min, but this time may not be enough to induce cybersickness. A previous systematic review reported that cybersickness appears approximately 10-15 min after VR immersion 62 . Fourth, subjects were applied VR while sitting on a chair without actual locomotion. Therefore, the cybersickness induced in this study may have different mechanism with that occurring in a more interactive experience during VR application. Despite these limitations, our study has the following strengths. To our knowledge, this is the first study investigating the adaptation effect of cybersickness when applying VR with different orders of exposure to VR videos with different degrees of shaking to participants. Although there was no adaptation effect observed in the results of this study, the application of VR in clinical settings will be further expanded as evidence is accumulated in future research. In addition, we identified clinical risk factors for cybersickness, including a comprehensive assessment of psychiatric, ophthalmological, and otologic assessments, as well as analysis of physiological parameters, which are promising biomarkers for psychiatric diagnosis.
The convergence of medicine and new technology is gradually expanding. In psychiatry, cybersickness is an important issue to overcome for interventions such as exposure therapy in PTSD or anxiety disorders and relaxation in a high stress population to be performed efficiently. Cybersickness can not only cause discomfort during the medical use of VR, but it can also make patients reluctant to use VR, thereby reducing its accessibility. In addition, although not replicated in this study, there is also evidence that cybersickness causes a change in heart rate, cutaneous thermoregulatory vascular tone, and prolongation of reaction time 63 . Another study found a change in brain perfusion during the experience of cybersickness 64 . These results suggest that cybersickness should also be handled in terms of safety, and clinicians should prepare for multiple scenarios before the application of new technologies.
In conclusion, the order of applying VR with different degrees of shaking did not affect to cybersickness. While smoking was a protective factor, more expression of affect was a risk factor for cybersickness.