Consequences of Virtual Reality Experience on Biomechanical Gait Parameters in Children with Cerebral Palsy: A Scoping Review

ABSTRACT Virtual reality (VR), coupled with motion tracking, can investigate walking in a controlled setting while applying various walking challenges. The purpose of this review was to summarize the evidence on consequences of VR on biomechanical gait parameters in children with cerebral palsy. MEDLINE, Embase and Web of Science were searched. Among 7.574 studies, screened by two independent reviewers, seven studies were included, analyzing treadmill (n = 6) or overground walking (n = 1) under VR. Most frequently reported were the spatiotemporal parameters walking speed, stride length, step width, stance phase, and the kinematic parameters range of knee flexion and peak ankle dorsiflexion. However, methodological approaches and reporting of the results were inconsistent among studies. This review reveals that VR can complement information gained from clinical gait analysis. However, this is still an emerging field of research and there is limited knowledge on the effect of VR on gait parameters, notably during overground walking.


Introduction
Cerebral palsy (CP) is one of the most common neurodevelopmental disorder in the early childhood with a prevalence of approximately 2-3 per 1000 births. 1,2It occurs due to nonprogressive injuries in the developing fetal or infant brain. 3mpairments in movement and postural control are often accompanied by psychological and social comorbidities in CP.The framework of the International Classification of Functioning, Disability and Health (ICF) provides a holistic approach to understand the multidimensional factors affecting the health and well-being of children with CP. 4,5 While functional and structural outcomes are extensively analyzed in the literature, less is known about personal and environmental factors. 6Being able to adapt walking to different environments and to explore these adaptations is crucial to overcome these barriers.
Biomechanical parameters are often used to quantify pathological gait and posture in these children, 7,8 referring to the ICF component of body function and structure.Laboratory-based three-dimensional (3D) motion capture systems are considered the gold standard in the collection of biomechanical data. 9hree types of data are traditionally collected with motion capture systems to quantitatively analyze human walking: kinematics, kinetics and electromyographic (EMG) data. 10The collected data is mainly used to verify the progress of specific walking impairments or treatment strategies to improve walking.Following the contextual factors of the ICF, it is important for clinical decision making to additionally consider the potential impact of environments that resemble real-life-scenarios on child's walking adaptations.Virtual reality (VR) can be used to simulate different scenarios of daily life.These scenarios are more complex than the artificial laboratory setting.They impose a challenge to walking and are difficult and potentially dangerous to implement in the real world.
VR is increasingly used as a tool to support rehabilitation of different neurological disorders, e.g., Parkinson's disease. 11ogether with enhanced functional motor skills, a case study in CP showed neuroplastic adaptations induced by a VR-based therapy. 12Additionally, VR increases compliance and motivation in these children as compared to conventional home exercises. 13These could be important determinants of enhancing motor skills and thus the effectiveness of treatment. 146][17] Recent studies have analyzed the potential of VR to assess postural control by quantifying center of pressure (COP) in children with CP. 18,19 While non-immersive VR experience is usually visualized on a screen in front of the user, 20 immersive VR can be accessed through a head-mounted display, providing the perception of being surrounded by the virtual environment.This perception of being physically present in a nonphysical world makes the laboratory environment disappear and with it the feeling of being obliged to show the "optimal gait pattern" (observation bias).The child might fall back into the personal gait pattern due to distraction by the virtual world.
A scoping review was conducted to systematically map the current evidence on immediate consequences of VR experience on biomechanical gait parameters collected via motion capture systems in children with CP and to identify existing research gaps.The review provides an overview of the VR conditions being used in the included studies to challenge gait and the type of biomechanical parameters being reported to quantify gait adaptations.The review highlights the potential of VR being used in the clinical gait analysis setting to assess and understand gait deviations in children with CP in virtual environments that resemble real-world scenarios and to assist clinical decision making.In the long term, a set of biomechanical gait parameters ("biomarkers") that describe the impact of VR experience on walking patterns, may support the use of VR as a diagnostic and outcome assessment in the context of the clinical gait analysis.

Study Design
We classified this paper as a scoping review due to the broader and more exploratory scope of the research question.The methodology follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) checklist. 21,22The methodology of this review is registered in PROSPERO (CRD42022313822).

Eligibility Criteria
The inclusion criteria were based on the framework of defining the Population, Exposure, Comparator, and Outcomes (PECO). 23opulation: Children with CP Exposure: VR experience Comparator: -Outcomes: Biomechanical gait parameters Included were studies published in English, French, Italian or German.We excluded studies using VR as an intervention tool, not combining VR simultaneously with gait analysis, and analyzing other movements than gait.Additionally, we excluded animal-only studies.

Information Sources
A comprehensive literature search was conducted on MEDLINE via PubMed, Embase and Web of Science.The search was performed from inception dates to the 16 th of December 2021.Additionally, reference lists of the included studies were screened manually.

Search Strategy
We developed the search strategy in consultation with a medical information specialist.The search term included text words and subject headings covering the eligibility criteria and was adapted to each database (Table A1).Text words and subject headings were combined using the Boolean operators AND and OR.NOT was applied to exclude animal studies using validated search filters for PubMed and Embase.

Selection Process
After deduplication in the citation manager Zotero, two reviewers (RL, MO) independently screened titles and abstracts for eligibility using the web-tool Rayyan.Any discrepancies in the screening results were resolved with a third author (EV).Full texts of the remaining studies were independently assessed by two reviewers (RL, MO).Discrepancies were again discussed with a third reviewer (EV).

Data Collection Process
Two reviewers (RL, MO) independently extracted data from the included studies.Extracted characteristics included author and year of publication, study location, study design and main objective, characteristics of participants, applied (baseline) walking and VR conditions, type of motion capture system used, biomechanical outcome parameters assessed under VR exposure and key findings.To obtain additional data, the authors of the included studies were contacted.

Risk of Bias Assessment
Following recommendations for quality assessment tools, 24 risk of bias was independently assessed by two reviewers (RL, MO) using the AXIS Critical Appraisal Tool for cross-sectional studies, 25 and the Joanna Briggs Institute (JBI) checklist for analytical cross-sectional studies. 26Discrepancies in assessments were discussed and resolved.None of the studies were excluded based on the risk of bias assessment.

Study Selection
The search retrieved 7.574 studies in total.After assessing titles and abstracts for relevance, 53 potentially eligible full texts were screened.We ended up with seven studies being included in this review (see Figure 1 for a detailed PRISMA flow diagram of the screening process).

Risk of Bias Assessment
According to the AXIS critical appraisal tool, the major sources of bias were lack of sample size justification, insufficient description of the overall population and lack of measures to address and categorize non-responders (Table 2).
The JBI checklist revealed the following main sources of bias: incomplete description of the population, incomplete description of the condition for inclusion and lack of analyzing potential confounding factors (Table 3).

Characteristics of VR Conditions and Systems
Table 1 provides a detailed summary of the explored walking and VR conditions in the included studies.8][29][30][31] The walking speeds during the VR walking conditions were either completely self-paced, 27,28 self-paced fixed [29][30][31] or completely fixed. 28The GRAIL studies compared different walking conditions with and without VR (Table 1).In all GRAIL-studies, conditions with and without VR were randomized and participants were fixed with a non-weight bearing safety harness during treadmill walking in the virtual environment to prevent falls.
The Computer-Assisted Rehabilitation Environment (CAREN) system, a six degrees-of-freedom movable platform, was used in one of the included studies 32 (Table 1).
The GAITRite system, a pressure sensitive walkway, was used in one of the included studies 33 (Table 1).

Type of Biomechanical Parameters
Table 4 summarizes the biomechanical parameters measured and reported in the included studies for the VR condition.Spatiotemporal parameters were most often reported, especially walking speed, stride length, step width and % stance phase.
Decreased walking speed was reported for level treadmill walking in a virtual environment, compared to walking outside the lab, 27 and for uphill treadmill walking in a virtual environment, compared to walking at ground level. 32Applying fast optic flow speed virtually led to an increased walking speed compared to normal or slow optic flow speed. 33o differences in walking speed were found when comparing treadmill walking with and without VR. 28ecreased stride length was reported for treadmill walking in a virtual environment, compared to natural walking outside the lab, 27 and for uphill treadmill walking, compared to walking at ground level. 32Applying virtual feedback on the hip angle during treadmill walking led to an increased stride length, compared to no feedback. 31One study found no difference in stride length between treadmill walking with and without VR. 28tep width remained unchanged when comparing treadmill walking with and without VR, 28 when applying virtual feedback either on the hip or knee angle, 31 or when walking on an inclined treadmill in a virtual environment. 32Virtual feedback on knee extension led to a wider step width compared to no feedback. 29ncreased stance phase was reported for treadmill walking in a virtual environment, compared to natural walking outside the lab, 27 and for uphill treadmill walking, compared to walking at ground level. 32No differences in stance phase were shown between treadmill walking with and without VR 28 or when applying virtual feedback. 29ecreased range of knee flexion was reported during treadmill walking in a virtual environment compared to  overground walking in a conventional gait lab. 27Virtual feedback on the knee angle led to an increased range of knee flexion in CP patients. 29,31No differences in knee flexion were found when comparing treadmill walking with and without VR. 28eak ankle dorsiflexion during stance and swing tend to decrease during treadmill walking in a virtual environment compared to overground walking in a conventional gait lab. 27Another study showed an increased peak ankle dorsiflexion during stance when applying virtual feedback on the hip angle, compared to no feedback. 31Peak ankle dorsiflexion remained unchanged when comparing treadmill walking with and without VR, 28 or when applying virtual feedback on knee extension, step length or ankle power 29 (see Table 1 for a summary of the studies' key findings).

Discussion
The aim of this review was to systematically map the existing evidence on immediate consequences of VR experience on biomechanical gait parameters collected via motion capture systems in children with CP.The number of studies was substantially reduced during the screening process.Finally, the review included seven studies that fulfilled the predefined eligibility criteria.
Self-selected fixed-speed treadmill walking in a VR environment combined with biofeedback led to increased hip and knee extension, 29,31 increased step length and increased ankle power generation at push-off. 29Uphill treadmill walking in a virtual environment decreased stride length, increased stance phase and decreased peak knee flexion during swing phase. 32ince these studies compared the same VR condition, either with and without biofeedback or with and without uphill treadmill walking, no conclusions can be regarding the impact of VR itself.Rather one can assume, that the biofeedback or the uphill treadmill condition itself led to changes in gait.Another study comparing treadmill walking with and without VR found no changes in walking speed, 28 which may be due to the fact that participants experienced walking in the VR environment similar to normal overground walking and hence did not need to adapt their gait patterns.However, walking in the VR environment increased absorbed ankle work and range of hip extension. 28Another study found that VR treadmill walking increased stride width, pelvic tilt and knee flexion at initial contact, compared to normal overground walking and walking outside the lab. 27The latter studies showed how the VR immersion itself led to changes in different gait parameters, which were mainly adapted toward gaining more stability during walking.Findings comparing treadmill walking with and without VR in healthy participants showed that changes in gait patterns may not only be related to the VR exposure itself but also to the treadmill speed mode. 34Hence, changes in gait patterns solely due to VR exposure itself are difficult to differentiate due to the application of a treadmill and the lack of a control condition including treadmill walking without VR exposure.Also, the changes in gait parameters are also likely related to the applied biofeedback.
The comparability among study results is limited due to the variability in participants' age, the applied VR walking conditions, sample sizes and the way gait outcomes were calculated and reported.
Compared to overground walking parameters in healthy controls, 35 all included studies reported decreased walking speed and stride length.All but one study reported increased stride width and all studies reported increased stance phase for the VR walking condition in children with CP.
Gait parameters investigated in studies on fixed speed treadmill walking 29,31 should also be interpreted with caution, although studies revealed similar biomechanical parameters between self-paced and fixed speed treadmill walking. 36ost of the included studies reported changes in biomechanical gait parameters on a dual-belt treadmill in a virtual environment.The GRAIL system with an integrated virtual environment was the most frequently used system.However, due to the semi-circular screen used to project the visual information in front of the participant, the method is less immersive, whereas the treadmill could be seen as a tool to increase immersion.The VR condition applied in the studies was most frequently combined with biofeedback.The studies reported changes of walking in a virtual environment on spatiotemporal, kinematic, and kinetic parameters, thus highlighting the opportunities of VR to assess gait adaptations in real-world scenarios by maintaining a safe laboratory setting at the same time.
The increased evidence on combining treadmill walking with VR emphasizes the advantage of treadmill systems that integrate VR and motion capture at the same time.This makes it possible to capture biomechanical parameters simultaneously to VR exposure.However, evidence on how virtual exposure changes these parameters during overground walking is still lacking.Previous studies revealed differences between overground and treadmill walking in spatiotemporal, kinematic, and kinetic parameters.Therefore, results from studies involving VR and treadmill walking may not be used to predict the effect during overground walking.
Type and amount of assessed gait parameters varied considerably among the included studies.Moreover, the calculation of these parameters is frequently not reported in enough detail to allow comparability between results.Initial contact and time-off can either be detected using vertical ground reaction forces with a predefined threshold 29 or based solely on kinematic data 27,31 using specific algorithms. 37Stride width can be calculated using either the mediolateral distance between left and right heel 27 or the distance between the medial malleoli.Both approaches lead to very different results when the child's feet point inwards instead of straight.Across the included studies, spatiotemporal, and kinematic parameters were most often reported, whereas kinetic and EMG data were less reported.Hence, a core set including at least specific spatiotemporal and kinematic gait parameters may be essential to assess gait patterns in children with CP.This may standardize and facilitate future assessments and monitoring of the treatment process.
The current review has several strengths and limitations.A limitation of this review is that the included studies provide a low level of evidence due to their cross-sectional design.This design limits the ability to make causal inferences from the observational data, meaning that the observed impact of VR on gait needs to be interpreted with caution.Sample sizes were small overall.Thus, selection bias may lead to a nonrepresentative sample of the overall population.A control group was lacking in most of the studies, indicating its more explorative character.Also, comparing the same walking condition with and without VR would increase the probability that the changes in gait were caused by the VR itself and not by other variables.The included studies varied considerably in types and combinations of gait outcome parameters, which limits the ability to draw conclusions on the consequences of VR on gait outcomes.Most of the studies came from the same working group in the Netherlands, assuming that the area of research, the knowledge and the technical equipment is still country focused.Another limitation is the use of treadmill walking in the included studies instead of overground walking.Thus, findings cannot be transferred since the biomechanical comparability between treadmill and overground walking is still questionable. 38Additionally, we cannot conclude that reported changes in gait parameters are related to the VR exposure alone and not to the treadmill walking.This is also underpinned by the fact that all but one study 33 did not compare walking with and without VR exposure using the same study setting.Applied VR environments in the included studies are often not described in sufficient detail to allow comparison between findings.Moreover, studies varied VR walking conditions by adding e.g., biofeedback features or varying the treadmill specifications (see Table 1 for a detailed summary on the general study characteristics).
The strength of this review is the systematic analysis of changes on biomechanical gait parameters due to VR exposure.In contrast to the huge number of studies analyzing treatment effects of VR on gait parameters, our review addresses more the potential of VR being used as an assessment tool for gait.Risk of bias of the included studies was analyzed using two different tools to cover different aspects that may lead to increased risk of bias.Another strength of this review is the focus on a well-defined population with a specific age range and diagnosis, which may facilitate the transferability of the review results to a comparable clinical context.
Future research should focus on how pure and immersive VR exposure changes overground walking patterns, since the treadmill itself may already disturb walking.Applying standardized virtual environments may help to detect gait impairments that may not be visible in the routine lab-based gait analysis.A standardized set of biomechanical outcome measures may reduce diversity in reported parameters, help to establish quality standards and support the use of VR in diagnosis and treatment of children with CP.
In conclusion, our review revealed that the current evidence on immediate VR consequences on gait parameters primarily focuses on treadmill walking.The VR conditions applied in the included studies differed considerably.Also, studies were mainly highlighting spatiotemporal findings, thus lacking the impact of VR on kinematic and kinetic parameters.The review showed that combining VR with clinical gait analysis is still an emerging field of research.There is limited knowledge on the effect of VR on gait parameters, notably during overground walking.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Figure 1 .
Figure 1.PRISMA flow diagram of the study selection process.

Table 1 .
Characteristics of studies included in the review.(s)CP=(Spastic) Cerebral Palsy; TD=Typical Development; VR=Virtual Reality; GRAIL=Gait Real-time Analysis Interactive Lab system; GMFCS=Gross Motor Function Classification Scale; EMG=Electromyography; CAREN=Computer-Assisted Rehabilitation Environment system.

Table 2 .
Risk of bias of the included studies based on the AXIS Critical Appraisal Tool for cross-sectional studies(25).Y=Yes; N=No; ND=not disclosed; NA=not applicable; UC=unclear.

Table 3 .
Risk of bias of the included studies based on the Joanna Briggs Institute (JBI) checklist for analytical cross-sectional studies(26).Y=Yes; N=No; NA=Not applicable.

Table 4 .
Biomechanical gait parameters measured in the included studies during virtual reality exposure.Parameters highlighted in yellow were reported in 3 studies, in orange in 4 or more studies.COM=Center of Mass; COP=Center of Pressure; IC=Initial Contact; LR=Loading Response.