Unraveling the effects of virtual reality overground walking on dynamic balance and postural control

This study analyzed the effects of walking freely in Virtual Reality (VR) compared to walking in the real-world on dynamic balance and postural control. For this purpose nine male and twelve female healthy participants underwent standard 3D gait analysis while walking randomly in a real laboratory and in a room-scale overground VR environment resembling the real laboratory. The VR was delivered to participants by a head-mounted-display which was operated wirelessly and calibrated to the real-world. Dynamic balance was assessed with three outcomes: the Margin of Stability (MOS) in the anteroposterior (AP-MOS) and mediolateral (ML-MOS) directions at initial-contact, the relationship between the mediolateral Center of Mass (COM) position and acceleration at mid-stance with subsequent step width, and trunk kinematics during the entire gait cycle. We observed increased mediolateral (ML) trunk linear velocity variability, an increased coupling of the COM position and acceleration with subsequent step width, and a decrease in AP-MOS while walking in VR, but no change in ML-MOS when walking in VR. We conclude that walking in VR results in a less reliable optical flow, indicated by increased mediolateral trunk kinematic variability, which seems to be compensated by the participants by slightly reweighing sensorimotor input and thereby consciously tightening the coupling between the COM and foot placement to avoid a loss of balance. Our results are particularly valuable for future developers who want to use VR to support gait analysis and rehabilitation.

Augmented, Mixed, and Virtual Reality (VR) technologies extend our reality by 2 merging the virtual with the real world thereby creating a fully immersive experience. 3 The global market size for these technologies was approximately worth 31 billion U.S. 4 dollars in 2021 and is projected to rise to almost 300 billion U.S. dollars by 2024 [1]. 5 These developments raise great expectations for future applications in various domains 6 such as entertainment, gaming, education, and fitness. However, the healthcare and 7 diagnostics sector is expected to be most advanced by these technological developments. 8 For example, VR has been and is still being intensively explored as a potential tool to 9 aid gait rehabilitation for both clinical practice and research [2,3]. 10 Thus, it is not surprising that there is a constantly growing body of literature 11 highlighting its potential for studying and aiding gait rehabilitation in patients with 12 December 12, 2022 1/16 various movement or balance disabilities such as Parkinson's Disease [4,5], Cerebral 13 Palsy [6], stroke [7][8][9], amputation [10], or aging [11][12][13][14]. While in the past VR has been 14 traditionally used on treadmills, advancements in off-the-shelf head-mounted-display 15 (HMD) technologies now allow to create immersive VR experiences where one can freely 16 walk around and navigate through virtual overground environments whereby one's 17 real-life motion is reflected in the VR environment. Immersive overground VR 18 environments offer partly unexplored possibilities for gait rehabilitation and research. 19 They can be used to train activities of daily living [8], obstacle crossing [15], or allow 20 the examination of effects for conditions which usually are difficult to evaluate such as 21 visual impairments, dual tasks, or various environmental effects like diffuse lighting 22 conditions or crowded places [16][17][18]. 23 However, before such room-scale VR environments can be used as a purposeful tool 24 in clinical and research practice it is important to understand if and to what extent that 25 technology impacts our individual gait pattern. To date, there is only limited and partly 26 inconsistent research available which specifically addressed this question for VR 27 room-scale applications. Briefly summarized, researchers found a decrease in stride 28 length [19], a decrease of walking speed, step length, and an increase of double support 29 [20], while others reported a decrease in cadence [21]. In addition, Janeh et al. [22] 30 reported that the decrease of walking speed during walking in the VR compared to the 31 real-world did not normalize even during prolonged exposure to the VR environment. In 32 contrast, one study found that a prolonged VR experience lead to a reduction in most 33 VR-related gait adjustments [19]. The study of Yamagami et al. [23] is one of the few 34 ones that assessed the effects of VR overground walking for people with Parkinson´s 35 disease. They found similar effects, such as reduced walking speed, step length, and 36 increased step width compared to the real-world laboratory. Interestingly, Martelli et 37 al. [19] and Yamagami et al. [23] also reported a clear increase in variability for step 38 width and length. In a previous study, our group examined the effects of a room-scale 39 VR environment on healthy adults by means of standard clinical 3D full-body gait 40 analysis [24]. The results of our study confirmed that individuals walk significantly 41 slower, along with an increased double support time as well as increased variability in 42 step width and foot off. Further, we found a markedly increased gait variability in the 43 lower-extremity gait kinematic patterns. Changes in kinematic and kinetic gait patterns 44 were small. Based on the currently available evidence it may be concluded that 45 individuals seem to adapt towards a more conservative or cautious gait. Research 46 demonstrates that individuals walking in challenging environments tend to adopt a 47 more cautious gait pattern to mitigate the risk for falling [25][26][27]. This might also be 48 the case for walking in VR room-scale environments. However, the simple analysis of 49 spatiotemporal, kinematic, and kinetic gait patterns that the literature currently offers 50 might not provide a complete answer to the question what the effects of VR are on 51 dynamic balance and postural control. 52 For this reason a more specific analysis is needed to better understand potential 53 effects of VR on our gait behavior. Gait is a dynamic motor task where the Center of 54 Mass (COM) is held outside of the base of support (BoS) for the majority of the gait 55 cycle, hereby defined as dynamic balance [28]. In this scenario, stability is achieved by 56 the neuromuscular system counteracting the gravitational and joint reaction forces 57 acting on the upper body to avert an uncontrolled fall [28]. One of the primary 58 mechanisms to counteract these forces and maintain dynamic balance is through 59 effective foot placement [28][29][30]. To achieve appropriate and effective foot placement, 60 the neuromuscular system predicts the future kinematic state of the COM to determine 61 upcoming foot placement at initial-contact [29]. In the anteroposterior direction, the 62 passive dynamics of the body are exploited requiring minimal guiding information from 63 automated subcortical and brain stem regions to determine foot placement [29,30]. However, in the mediolateral direction, appropriate foot placement requires active 65 sensory information processing from higher-level cortical regions [29]. Specifically, in 66 this direction, the visual and vestibular systems provide information about head 67 orientation which is then integrated with proprioceptive input from the trunk [29,30].

68
The culmination of this process provides the neuromuscular system with an estimate of 69 the dynamical state (position, velocity, and acceleration) of the COM which it then 70 utilizes to determine upcoming mediolateral foot placement [29]. Indeed, previous 71 research demonstrates that upcoming foot placement is determined by the kinematic 72 state of the COM during the preceding contralateral mid-stance [29,[31][32][33]. As walking 73 is a time-varying task with equally time-varying COM dynamical states, a constant 74 sensory reweighing occurs to achieve foot placement [29].  [24]. That information could be of great value in assessing the 83 potential fall risk individuals might be exposed to in VR room-scale environments. In   The latter two were not relevant for this study.

99
Virtual Reality Room-Scale Environment

100
To immerse participants in the room-scale overground VR environment they wore a 101 HTC Vive Pro HMD which was operated wirelessly and calibrated to the real-world.

102
One HTC Vive 2.0 tracker was strapped to each foot to track and display the feet in VR 103 in real-time (Fig 1). This allowed volunteers to have a visual indication of the current 104 position in the VRLab and their body posture while navigating through the VR CH) was used for collecting ground reaction force data at 300Hz. To collect participants' 111 kinematics, the extended Cleveland Clinic marker set [34] was used for the lower (1) where vCOM is the velocity of the COM, g is the gravitational acceleration and l is 138 the height of the inverted pendulum which was determined as the distance of the forward, which is the goal of forward walking. Since velocity is taken into account in the 152 xCOM , the AP-MOS will lie outside the BoS during IC.

153
To assess the ability of the dynamical state of the trunk to determine mediolateral 154 foot placement, we used similar to [31] a multiple linear regression model to relate the 155 trunk COM position and its acceleration at mid-stance with the subsequent step width. 156 Mid-stance was defined as the time point when the frontal velocity of the trunk COM 157 was zero. The COM position was defined as the horizontal distance between the vertical 158 projection of the COM on the ground and the mid-point of the line connecting the heel 159 and toe markers of the stance foot at mid-stance.
Step width was calculated as the 160 frontal plane distance of the mid-point between toe-and heel-markers for both feet 161 during two consecutive steps (see Fig 2). For the regression analysis, the trunk COM 162 positions and accelerations at mid-stance for each left and right step were paired with 163 the following step widths for each participant and lumped together in one sample. This 164 resulted in a regression equation as followed for each walking condition. The explained 165 variance in terms of the coefficient of determination (R 2 ) served as outcome.
Lastly, to assess postural control we used the AP, ML, and vertical kinematics of the 167 trunk to understand the impact of the VR on trunk kinematics. The T10-marker, 168 placed right above the tenth thoracic vertebra, served as a surrogate to describe trunk 169 motion. We have calculated the velocity profiles as well as the variability in terms of its 170 standard deviation for each individual across the entire gait cycle and averaged them

198
A Shapiro-Wilk test indicated a normal distribution for the AP-MOS but a non-normal 199 distribution for ML-MOS. A dependent t-test identified a significant difference in   These findings are significant as instability and fall risk is greater in the mediolateral 234 compared to the anterposterior direction while walking [28][29][30].

235
The increase in mediolateral trunk linear velocity variability may have stemmed 236 from the difference in the visual scene between the VR and the real-world. Previous 237 research demonstrates that postural control in the mediolateral direction, compared to 238 the anteroposterior, requires the effective processing of visual information [29,40]. Higher-level cortical structures integrate this input with somatosensory and vestibular 240 input to provide information regarding the orientation and movement of the head 241 relative to the trunk [29,40,41]. In turn, this provides the neuromuscular system with 242 an estimation of COM kinematics during walking [29]. In our study, the increased 243 mediolateral trunk velocity variability may indicate that participants had difficulty in  [30,45]. However, in our preceding study [24] on this sample, we observed a 269 simultaneous increase in participants' step width variability. Although increased 270 variability is a marker for falls in older adults [30,46,47], current evidence suggests that 271 increases in spatiotemporal variability indicate foot placement adjustment to support a 272 destabilized upper body in healthy young to middle-aged adults [30,45]. Adjusting foot 273 placement, as reflected by increased spatiotemporal variability, would account for the 274 lack of findings in ML-MOS as participants modified their base of support to maintain 275 their pre-existing level of dynamic stability in this direction. This is further 276 strengthened by the fact that participants also displayed elements of a "cautious gait" 277 strategy (slower walking speed, longer stride time, and increased double support time). 278 When dynamic stability and postural control is threatened, individuals take slower steps 279 to provide additional time for the COM to transition from the unloading to the loading 280 leg during double-support [48,49]. Additionally, the cautious gait strategy would 281 account for the reduced AP-MOS observed during the VR condition compared to the 282 real-world condition. Although vision is not closely linked to stability in the 283 anteroposterior direction, [50] suggested that individuals implement this strategy to 284 reduce the distance of the dynamical state of the COM to the base of support. As the 285 dynamical state of the COM is located ahead of the base of support during steady-state 286 walking, reducing the distance of the AP-MOS would facilitate the ability for 287 individuals to return it within the base of support during a potential loss of balance 288 such as when encountering an external perturbation. Thus, in our study, participants 289 may have adopted a slower walking speed and a cautious gait strategy to reduce the 290 likelihood of a balance loss in the event they encountered a potential perturbation in the 291 real-world while wearing the HMD.

292
The ability to maintain a pre-existing level of mediolateral dynamic stability despite 293 the altered optical flow in the VR may be due to sensorimotor reweighing which 294 occurred. In a literature review, [29] discussed that during walking, multisensory 295 information is integrated as a weighted average that is determined by the reliability of 296 each individual sensory component. As such, when a specific component (vision) 297 becomes less reliable, the sensorimotor system reweighs the afferent input to rely more 298 heavily on alternative sensory information (somatosensory and vestibular). This 299 reweighing may explain not only the lack of changes in ML-MOS, but also the increased 300 R 2 values during the VR condition. Indeed, we observed that COM kinematics 301 (position and acceleration) accounted for 64 percent of the variance in step width during 302 the VR condition compared to 54 percent during real-world walking. Hurt et al. [31] 303 proposed that a stronger relationship between upper body kinematics and step width 304 may indicate increased voluntary control to maintain dynamic balance while walking.

305
As such, in our study, the potential sensorimotor reweighing that occurred, due to the 306 altered optical flow in the VR, may have caused participants to consciously tighten the 307 coupling between their COM and foot placement to avoid a loss of balance. However, it 308 is unclear whether the change in magnitude of 10 percent, compared to a relative 309 change of either 15 or 20 percent, holds a direct implication for dynamic balance. 310 Interestingly, Hurt et al. [31] reported differences of approximately 9% between young 311 (24.5 ± 3.3 years) and older adults (60.6 ± 5.6 years). This might provide an 312 approximate indication as older adults have reduced dynamic balance compared to 313 healthy young adults [30]. [31], therefore, suggested this increase in the older adults was 314 indicative of a more active strategy to control gait. However, they used a treadmill and 315 the extent to which the relationship between variations in step width and COM state 316 would be altered treadmill versus overground is presently not known. As such, future 317 research should examine whether relative changes in magnitude affect an individual's 318 dynamic stability level as well as if treadmills further influence the relationship between 319 step width and the COM state. Recent research indicates that VR based exergaming 320 can effectively increase balance and reduce fear of falling in elderly [51][52][53]. VR is also 321 increasingly used in patients with neurological disorders, such as Parkinson Disease. In 322 a systematic review and meta-analysis, [54] recently confirmed that VR assisted balance 323 training is highly effective in improving balance in patients with Parkinson's Disease. 324 While to date, the majority of research only uses non-immersive VR systems such as 325 Microsoft's xBox and its wireless Kinect tracking system or non-immersive VR 326 combined with treadmill training, the speed at which immersive VR technologies are 327 currently developing suggests that there might be applications for fall-prevention and 328 functional training in near future where highly immersive VR could play an important 329 role. However, before such immersive VR technologies can serve as purposeful tools we 330 need to fully understand the impact they have on our gait behavior and postural 331 control. Unfortunately, as our study sample only comprised healthy individuals aged 332 between 21 and 56 years, our results are limited in their generalizeability to the elderly 333 population or patients with neurological disorders. The study by Yamagami et al. [23] is 334 the only study we are aware of which evaluated the effect of VR on gait characteristics 335 in patients with Parkinson Disease. They investigated whether Freezing-of-Gait (FoG) 336 provoking VR environments exacerbate gait impairments associated with FoG compared 337 to unobstructed VR and the physical laboratory. They found that walking speed was  Lastly, our results need to be interpreted with caution when being used to inform the 345 planning or development of prolonged exercise sessions using immersive VR as we do not 346 know if the observed effects cease over a prolonged use or are independent by usage time 347 and experience with VR. This is an important question which needs further attention. 348

349
Recent studies in general observed a consistent pattern of gait adjustments when 350 walking in VR overground environments compared to walking in reality [19-21, 23, 24]. 351 Most frequently reported effects are reduced walking speed, increased gait variability, 352 and step width which all point towards adjustments to a more cautious gait. Our study 353 further underscores this idea, and is the first to provide an explanation from the delivered to the participants with a HMD results in a altered optical flow, indicated by 356 increased mediolateral trunk kinematic variability, which seems to be compensated by 357 the participants by slightly reweighing sensorimotor input. Subsequently, participants 358 consciously tightened the coupling between their COM and foot placement to maintain 359 their already existing level of mediolateral dynamic stability. Although our results show 360 some adjustments in dynamic stability and postural control, these should not be 361 overestimated as we already showed that overall effects on the gait kinematic and 362 kinetic patterns are rather small [24]. Immersive VR is a rapidly developing technology 363 and it is reasonable to assume that VR and HMDs will become even more immersive in 364 the near future thereby further reducing the effects they currently have on gait stability. 365 Our study should thus be repeated in the foreseeable future with updated hardware and 366 with various patient groups to support its application as a purposeful tool in healthcare. 367

368
This study involving human participants was approved by the local ethics committee 369 (GS1-EK-4/682-2020) and was performed in accordance with the relevant guidelines and 370 regulations. All participants were informed prior to the study and gave written 371 informed consent.