Virtual reality as a telerehabilitation strategy for people with autism spectrum disorder during the COVID-19 quarantine scenario: physical activity, motor performance and enjoyment

Abstract Purpose People with autism spectrum disorder could benefit from physical activity during the pandemic and COVID-19 restrictions, mainly to maintain adequate physical activity. We aimed to evaluate the feasibility, enjoyment, and potential effect of telerehabilitation using a serious game named ‘MoveHero’. Materials and methods Registered in Clinical Trials (NCT04402034). We adopted a remotely run Telerehabilitation research design with 44 participants recruited: 22 People with ASD people and 22 non-ASD individuals. Results All participants safely participated, 100% adherence to sessions, ∼60% enjoying the task, and significantly improved performance, with better performance for the NA group at most practice moments. Conclusions Our findings support both how to implement a gaming intervention and the need to investigate the efficacy of serious games to motivate moderate intensity physical activity in people with ASD.


Introduction
Autism spectrum disorder (ASD) is characterised by a quantitative deficit in communication and social interaction, along with a series of restricted, repetitive, and stereotyped behaviours and interests [1].In addition to sensory reactivity, children with ASD have difficulties in visual praxis, somatodyspraxia, and vestibular skills [2].Characteristics such as changes in muscle tone, fine and gross motor functions, balance, and motor planning are also described with all these factors affecting computer use [3].
Moreover, people with ASD show significative impairments in motor performance and hand motor activities [4] and frequently present a decline in physical activity (PA) level due to a lack of motivation [5], and low interest and enjoyment [6].
Physical activity presents positive effects for people with ASD in sleep quality and cognition [7], and social interaction and communication skills [8], reducing the number of episodes of stereotypical behaviours [9], locomotor skills and muscular strength and endurance [10][11][12][13].Moreover, important factors such as fun, joy, tranquillity, satisfaction, and participation should be considered in all forms of PA for ASD individuals [14].
The impact of the outbreak of COVID-19 as a global pandemic [15] on social distancing, restricting mass gatherings, and encouraging people to stay at home in quarantine [16] has according to the parents of people with ASD been especially challenging and required more commitment than before, with people with ASD being particularly vulnerable to disorganisation from routine interruptions [17].It is well known that the capability of ASD individuals to be self-sufficient (in functions such as communication, socialisation, daily living, and motor skills) was crucially impaired by the COVID-19 pandemic [18,19].According to Gregor et al. [20] parents described that the amount of structure within different programs, ratio of staff to participants, location of the activity (indoor/outdoor), use of routines and schedules, acoustics and lighting (sensory inputs) of the space, and type of sport or activity, all influenced PA for ASD Individuals.
Thus, as people with ASD had a high probability of worsening their symptoms [21], interventions through Home-based telerehabilitation (HBTR) could be the most viable possibility to maintain socialisation and communication between the rehabilitation team and patients and could provide physical activity to individuals with disabilities that have parents or caregivers who are not able to travel to a clinic-centre.
HBTR can be defined as a rehabilitation modality in which physicians offer rehabilitation to individuals using telecommunication devices [22].HBTR can be as effective as conventional rehabilitation in improving activities of daily living, and enhances adherence to rehabilitation training [23,24].This modality can also make it possible to reduce the therapist and client time, allowing for more regular practice, in addition to facilitating access to on-site rehabilitation centres for those who are unable to travel [25,26].In ASD, the use of HBTR has been gaining traction globally for the past few decades, bringing positive results in occupational therapy [27], speech-language therapies [28], and approaches for social interactions between the patient, therapist, and their family [29].
Considering different possibilities for organising a HBTR program, a new approach that has been growing in popularity is the use of Virtual Reality (VR) games.VR has been used to elicit light, moderate, and vigorous levels of activity in some patients [30] and has great potential for engaging individuals in physical activity practice [31].People with ASD are able to use different interfaces and improve performance in a virtual task obtaining significant improvements in physical fitness, executive function, self-perception, and participation [32,33].
The use of VR systems is increasing in people with ASD [34], with promising improvements shown in both simple skills, such as reaction time [35], and more complex real-life skills, such as pedestrian skills, with higher involvement and motivation and positive transference from the virtual environment to real-life behaviours [36].According to Karami et al. [37] in a systematic meta-analysis about VR in individuals with ASD, VR technology can be a viable tool for designing interventions aimed at enhancing and improving different skills in people with ASD at any age, with a strong effect observed for daily living skills and a moderate effect for social, communication, and cognitive skills.
Despite these benefits, the use of VR games to improve physical activity in HBTR is still unknown.Thus, the objective of the current study was to test the feasibility of running a trial remotely, during the quarantine period (due to COVID-19), using a VR software platform created with a game that enables the rehabilitation team to manage a physical activity plan, considering the difficulties presented by people with ASD.We hypothesised that all participants would be able to participate safely in the trial, improve performance with the game practice, as well as increased physical activity and enjoyment.However, considering the difficulties that characterise individuals with ASD, we assumed these improvements and benefits would be more evident for the non-ASD group.If this hypothesis is confirmed, the results of this study will be relevant for the use of telerehabilitation as a complementary possibility to provide improvement in performance and physical activities to individuals with ASD.

Study design
Using a cross-sectional design, focused on quantitative data, this study was conducted between March 2020 and July 2020 (quarantine period defined by the state of São Paulo to reduce the transmission of COVID-19) and extended until February 2021.This study follows the guidelines and recommendations of the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) [38] and was registered in Clinical Trials (NCT04402034).In accordance with the Declaration of Helsinki (1964), later revisions, and the Research Governance Framework for Health and Social Care [39].

Participant recruitment
The researchers contacted 35 families with children with ASD who usually participated in a rehabilitation program at Fisioterapia Pedi atrica e Cia (a rehabilitation clinic located in a city in the countryside of São Paulo).After the first contact, during which an explanation of the activity was given, 13 families were unable to participate (10 without a computer or internet connection, 1 without the computer capacity to play the game, and 2 parents without time to help the individuals with ASD.Thus, 22 families agreed to participate in the study.After completion of the intervention by the individuals with ASD, we evaluated 22 non-ASD individuals matched by age and sex with the experimental group.The non-ASD individuals were recruited through disclosure on social media.If they wanted to participate, they got in touch with the researchers and scheduled a day and time to meet online (Figure 1).

Inclusion criteria
Young people with ASD were included if they: (1) had a clinical diagnosis of ASD, performed by a neuropediatric doctor or child psychiatrist, (2) were aged from 5 to 13 years, (3) had a light/moderate score on the CARS Scale [40], (4) did not use medication that could interfere with the studied variables, such as cardiac beta-blockers, (5) had access to a computer with a webcam, (6) had an internet connection to use the game, and (7) the individuals assented, and their legal guardians consented to participate in the research.Individuals with low vision or blindness and those who do not have technological devices to perform telerehabilitation were not included in the study.

Exclusion criteria
Participants were excluded if (1) they did not understand the tasks (comprehension of the task was assessed through 2 min of practicing the task, each match lasted 5 min, if the patient could complete the match without missing more than 10 spheres continuously, they were allowed to move on to the next match; (2) they had difficulties in maintaining attention and self-regulation that prevented the completion of virtual tasks; (3) they do not have appropriate devices to perform the virtual task (computer) or to contact the researcher (cell phone or other computer); (4) the system did not work during practice due to some technological failure (such as internet connection failure).

Instrument
In the current study, we used a platform called MoveHero (Figure 2), available for free use in Portuguese https://movehero.com.br/ and English https://movehero.com.br/en/.As presented by Martins et al. [41], MoveHero is a coincident timing task and consists of several spheres falling down the computer screen, with a musical rhythm to increase engagement.The participant is positioned in front of a computer and when the game starts the webcam captures the participant's movements and a representation of the player appears on the computer screen as an avatar (for more information about the game see Dias et al. [42,43], Silveira et al. [44], Moraes et al. [45] and Ribeiro [46]).

Assessments
To characterise the participants, we used: 1. the Childhood Autism Rating Scale -CARS [40] (parent-proxy report), to assess the severity of the autism, classified as follows: a total score of 15-29.5 was considered 'non-autistic,' a score of 30-36.5 was considered 'mild to moderate' autism, and a score of 37-60 was considered 'moderate to severe' autism [47].This scale presented internal consistency and sensitivity, and should be used along with other confirmatory tools [48].2. Sociodemographic information (parent-proxy report) was collected through a questionnaire, used to assess information such as sex, if they had feelings of loneliness, trouble sleeping, if the participant practiced physical activity, and the use of games in their routine.To analyse the Outcome, we used: 'not fun at all,' 1 'boring,' 2 'a little bit fun,' 3 'fun,' and 4 'great fun') was applied after the telerehabilitation, to verify the participant's level of enjoyment while performing the activity.This scale was developed to evaluate how people feel when interacting with the proposed non-immersive VR games and was previously used in other studies [53].3. The Rating of Perceived Exertion -CR10 (RPE) [54] (participant report), is a tool for monitoring the intensity of physical effort.The participant is required to score their feeling of tiredness from 0 to 10.This scale is used both in areas of high-performance sports and physical rehabilitation to monitor the changes caused by physical exercise in the cardiorespiratory, metabolic, and neuromuscular systems [55,56].

Procedure and design
The guardians of the participants were contacted by phone and asked to help the individuals with ASD to perform telerehabilitation sessions at home, led by the researcher over the phone, with a video call.During the video call, all the instructions on how to conduct the game and tasks were given to the parents and guardians.The guardians expressing an interest in the study received a link with the assent and consent forms to fill out and to complete a Google Forms survey for all measures.

Protocol
The protocol was divided into 4 matches where M0 was considered the first contact with the task (first day), when the participants gained understanding of the game (familiarisation match) by playing for 3 min.After the familiarisation and on a second day all participants answered the BRUMS and played matches M1, M2, and M3 (all three matches consisted of 3 min playing the game) with intervals of approximately 20 s between matches to receive the score and answer the RPE scale.After the 3 matches the BRUMS was answered again and the enjoyment scale was applied (Figure 3).
To determine the sample size, we used statistical software (G*Power 3.1.5)on the main outcome measure (i.e., the motor score).This calculation was based on data from five patients (pilot study).The power was 0.80; the a was 0.05, and the effect size was 0.65 (Cohen d).The sample estimation indicated that 40 participants would be necessary (i.e., 20 per group), and with an adjustment to allow for a withdrawal rate (20%), we aimed to recruit 24 participants per group, considering the intrinsic needs of the research, such as internet connection, computer capacity to play the game, and parents with time to help the participants with ASD, we contacted 35 families of individuals with ASD, to get 22 participants with ASD.

Community involvement statement
The researchers recruited an advisory board that comprised people with ASD, parents of children with ASD, and two clinical practitioners, all related to the clinic Fisioterapia Pedi atrica e Cia, a rehabilitation clinic located in a city in the countryside of São Paulo.

Data Analysis
The MoveHero game is considered a timing coincident task that provides a dependent variable with timing error (i.e., timing error was defined as the difference between the time when the sphere started dropping and the time the individual managed to hit the target with the avatar's hand).Thus, using MoveHero we can verify two important movement errors: 1) absolute error (AE), which demonstrates the accuracy of the movement, calculated by taking the absolute value of each raw score and disregarding whether the response was early or late, and 2) variable error (VE), which identifies the precision of the movement, calculated as the timing error resulting from within-subject variability.The calculation considered the square root of the sum of the square of differences between each score and the individual constant error mean divided by the number of trials (for details about these errors, see de Mello Monteiro et al. [57] and Moraes et al. [33]).
The absolute error (AE), variable error (VE), and percentage of hits and misses were submitted to a Multiple Analysis of Variance with factors: 2 (Groups: autistic and non-autistic) by 4 (Matches: M0, M1, M2, and M3), by 4 (Target Positions: P1, P2, P3, and P4) with repeated measures for Matches and Target Positions, as well as Physical Activity as a covariate (MANCOVA).Rating of Perceived Exertion (RPE) was submitted to Analysis of Variance (ANOVA); with factors 2 (Groups: autistic and non-autistic) by 4 (Moments: Mo0, Mo1, Mo2, and Mo3).The Enjoyment Scale was submitted to a chisquare test using the variables Groups (2: autistic and non-autistic) and how fun the game was (2: Less fun, More fun).Brunel Mood Scale data of 6 dimensions (Anger, Confusion, Depression, Fatigue, Tension, Vigour) were submitted to a paired samples t-test.The Least Significant Difference (LSD) was used as a post-hoc test.The partial eta-squared (gp 2 ) was reported to measure the effect size and interpreted as small (effect size> 0.01), medium (effect size> 0.06), or large (effect size> 0.14) [58].We also reported the Observed power (op).Values of p < 0.05 were considered significant.The statistical package used was the Statistical Package for Social Sciences (SPSS; IBM, Chicago, Illinois, USA), version 26.0.

Results
A total of 44 individuals participated in this study (22 individuals with ASD; 22 non-ASD).The mean age of the participants with ASD was 8.05 ± 2.3 (5 to 13 years old) and non-ASD was 8.95 ± 2.4, with three girls (13.5%) and 19 boys (86.5%) in each group.Regarding the CARS, the mean score of the ASD group was 32.8 ± 2.3.Data about participant's characteristics are presented in Table 1.
As we found a significant difference between groups regarding physical activity, we first ran the MANCOVA, using physical activity as a covariate, but we did not find any effect for this factor.Thus, we proceeded to the results taking physical activity as a covariate.For Absolute and Variable errors, the MANOVA found a significant effect for Matches [Wilks' lambda ¼ 0.295, F 6, 13 ¼ 5.18, p ¼ 0.006, g p 2 ¼ 0.71, op ¼ 0.93] with no further effects or interactions.The separate ANOVAs are described in the sections below:

Absolute error (AE)
There were significant main effects for Matches [F 3, 54 ¼ 14.Regarding the Target Position, although there was no significant interaction between Groups and Target Position, the post hoc test showed that participants from the ASD group presented better performance in the lateral targets (P1 ¼ 1235 ms and P4 ¼ 1087 ms; against P2 ¼ 1733 ms and P3 ¼ 1447 ms), while the non-ASD group presented better performance in the right targets (P3 ¼ 988 ms and P4 ¼ 770 ms; against P1 ¼ 1142 ms and P2 ¼ 1077 ms).The mean and standard errors are depicted in Figure 4.
Regarding the interaction between Group, Matches, and Target Position, the post hoc test showed that for the ASD group the improvement from M0 to M1 only occurred for Positions 2 (p ¼ 0.009) and 3 (p ¼ 0.009), with no significant improvements from M0 to M2, and from M0 to M3 the improvement was significant for Target Positions 2 (p ¼ 0.009) and 4 (p ¼ 0.005).For the non-ASD group, there were no significant improvements between any matches (Figure 4).

Percentage of hits and misses
Regarding Hits, there were significant main effects for Matches [F The post hoc test showed that both groups presented significant improvements in the percentage of hits from M0 to all other Matches (M1: p < 0.001; M2: p < 0.001; M3: p < 0.001).The post hoc for the interaction between Groups and Matches showed that the differences between groups were present in M1 (p ¼ 0.040), M2 (p ¼ 0.066) and M3 (p ¼ 0.006), but not in M0.
Considering Misses, only a main effect for Matches [F 3, 54 ¼ 1.97, p < 0.001, g p 2 ¼ 0.39, op ¼ 1.0] was found, and similarly to Hits, the post hoc test showed that both groups significantly decreased the percentage of misses from M0 to all other Matches (M1: p < 0.001; M2: p < 0.001; M3: p < 0.001).The means and standard errors are presented in Figure 5.
Regarding their enjoyment (Figure 6b), a significantly higher number of participants from both ASD and non-ASD groups  found the intervention more fun than less fun (Pearson chisquare: p ¼ 0.015)

Brunel Mood Scale (BRUMS)
The data on mood before and after the intervention are described in Table 2.

Discussion
The present study investigated the possibility of running a trial in telerehabilitation using a VR platform and found it possible for two-thirds of invited families to participate and for the intervention to be safe and feasible for participating children with ASD.We found that all participants presented improved performance on repeated trials, with better performance for the non-ASD group at most practice moments.Importantly, most participants (60%) enjoyed the games, and repeated exposures to the games demonstrated very high levels of enjoyment, with all participants completing all sessions, and exercising at exercise intensities reflective of moderate intensities at the later sessions.
We were able to observe factors that made the game easier or harder for people with ASD, such as the position of the targets.Taken together, our findings are exciting and support both how to implement a gaming intervention and the need for a remote trial to investigate the efficacy of serious games to motivate moderate intensity physical activity over the long term in people with ASD.Some important findings are presented below.

Performance improvement
The individuals with ASD presented similar movement accuracy (absolute error) and precision (variable error) to non-ASD individuals at the familiarisation moment in all positions (M0first day).After the familiarisation, both groups showed improvement in performance during the game considering absolute error, however, the non-ASD group performed better.This result was expected as studies using VR in individuals with ASD identified improvement in motor skills for them, as well as better performance for non-ASD groups [33,35], likely due to altered sensorimotor control and a greater reliance on proprioceptive and tactile information in individuals with ASD [59,60].
Another novel observation in performance was that the ASD group had a higher performance (i.e., they hit a larger number of spheres) on the external targets of the game with the same performance as the non-ASD group in absolute error.One factor that may have contributed to this result is the common excessive activity in individuals with ASD and the difficulty in reaching or crossing the middle line, which could facilitate distal practice (i.e., distal movements have benefits with more range of movement) [61,62].

Physical activity (PA)
We believe that the most important result from our study was the increase in the rating of perceived exertion (RPE scale) continuously during the practice in the VR game for both groups where intensity levels were moderate for most in later sessions.Suggesting gaming is an effective approach to motivate people with ASD to exercise at moderate levels and possibly achieve health and well-being benefits.
Before the pandemic, the use of online technologies was not a usual practice among children [7], while now, children and their parents or caregivers are more receptive to this kind of practice, especially those who have limited mobility or live far from the places they used to go to receive treatment.Sorensen & Zarett [63] in a review about the benefits of physical activity found strong evidence that physical activity can be effective in clinical motor signs presented in individuals with ASD.We agree with the statement from Cynthia et al. [64] that enhancing motor coordination and motor skills may not only benefit PA participation, but also translate to other aspects of daily living, such as social skills and manual dexterity.Thus, our results can be considered encouraging and may positively influence rehabilitation professionals prescribing telerehabilitation to people with Autism Spectrum Disorder, providing an effective alternative to promote PA, since autistic individuals with ASD prefer videogames for recreational activity, and tend to play with high frequency.

Enjoyment
In the pandemic scenario, parents of people with ASD reported that the anxiety, fear, stress, low mood and difficulty in emotion management were symptoms that their children were facing during the isolation time [65,66].With the possibility of changing their humour by giving a protocol that could promote motivation, another important result of our study, which may be responsible for the improvement in performance and increase in PA was the impact of enjoyment using the VR game during HBTR.The majority of ASD group (59.5%) and non-ASD group participants (91%) characterised the task as 'fun' or 'very fun'.
Tse et al. [7] demonstrates that children with ASD struggle with self-regulation of anger that can be considered a characteristic of people with ASD when practicing PA.Individuals with ASD presented reduced tension and increased vigour after practice with the MoveHero game, which could suggest that this platform provides the expected positive changes in mood from physical activity, although we also found increases in anger and fatigue after the practice.Therefore, considering the decline in PA levels in ASD related to factors such as lack of motivation, low interest, and low enjoyment [5,6,45,67,68], our results can be considered evidence of an interesting tool that facilitates the use of technology to promote PA, making it attractive and enjoyable for individuals with ASD.

Limitations
Although we found interesting results, we can point out some limitations of the present study: (1) this was a one-day protocol and a protocol with a training program should provide important information and continuous benefits to individuals with ASD; (2) we did not analyse specific motor and cognitive abilities because they are Legend: Table 2 represents the comparison between groups and moment of assessment (pre and post intervention) of mood using the Brunel Mood Scale.A value of p < 0.05 is considered significant.
difficult to assess in a telerehabilitation protocol; (3) we did not analyse the type of music used (which could influence the performance of ASD group); and (4) we should have measured the influence of the therapist and family members during the practice, as these data could be important to define future interventions;

Figure 2 .
Figure 2. Participant Positioning and game design.The task was to intercept different falling spheres using upper limb wave movements at the exact moment the spheres reached their specific target.Bottom right image, miss feedback -above this one, hit feedback.The game is available for free use in Portuguese https://movehero.com.br/ and English https://movehero.com.br/en/.
1. the Brunel Mood Scale -BRUMS[49,50] (participant report), this scale contains 24 simple mood indicators, such as feelings of anger, disposition, nervousness, and dissatisfaction that are noticeable by the individual being assessed.The possible score is 5 points according to how they feel about these sensations (0 ¼ nothing to 4 ¼ extreme).The scale takes about 1 to 2 min to complete.The 24 indicators of the scale comprise six subscales: anger, confusion, depression, fatigue, tension, and vigour [51,52].2. Enjoyment Scale (participant report), using smiley faces (0

Figure 4 .
Figure 4. Absolute error and variable error mean and standard error for both ASD and non-ASD groups.

Figure 5 .
Figure 5. Mean and standard error of percentage of hits, misses and the continuous score, for both ASD and non-ASD groups.

Figure 6 .
Figure 6.Mean and standard error of RPE score (a) and representation of the answers of the individuals regarding their enjoyment with the game (b) for both ASD and non-ASD groups.

Figueiredo
Santos de Almeida -Master's student in the Postgraduate Program in Health Promotion at the University of Franca -UNIFRAN; Talita Dias da Silva -Post-doctorate completed in Oxford/UK (2018) and ongoing in the Rehabilitation Sciences Program of the Faculty of Medicine of USP (with PNPD-CAPES scholarship in 2019); Ibis Ariana Peña de Moraes -PhD in Rehabilitation Sciences at the Faculty of Medicine of the University of São Paulo, with an internship abroad (sandwich doctorate) at the College of Medicine and Health of the University of Exeter -Exeter/UK; Lilian Del Ciello de Menezes -PhD in Medicine (Cardiology) Federal University of São Paulo, UNIFESP, Brazil.; Eduardo Dati Dias -PhD student in Rehabilitation Sciences at the Faculty of Medicine of the University of São Paulo; Luciano Vieira de Ara ujo -PhD in Bioinformatics from the University of São Paulo; Carlos Bandeira de Mello Monteiro -Postdoctoral fellow at the Department of Maternal and Child Health, Faculty of Public Health, University of São Paulo; Helen Dawes; Amanda Orasmo Simcsik -PhD student in Rehabilitation Sciences at the Faculty of Medicine of the University of São Paulo; Camila Aparecida de Oliveira Alberissi -Master of Science in the Physical Activity Sciences Program at the School of Arts, Sciences and Humanities of the University of São Paulo; Victoria Yanara Hernandes da Silva -Master in Cardiology Medicine from the Paulista School of Medicine of the Federal University of São Paulo -UNIFESP -EPM; Marisa Afonso Andrade Brunherotti -Post-Doctorate at the Department of Social Medicine of the Faculty of Medicine of Ribeirão Preto, University of São Paulo; Maria Georgina Marques Tonello -Post-Doctorate at the Polytechnic University of Madrid -Spain

Table 2 .
Brunel Mood Scale (BRUMS) for both groups, before and after the intervention.