Virtual reality tool for balance training

dezembro de 2021 Diana Raquel Barroso Gonçalves Rito Virtual reality tool for balance training dezembro de 2021 dezembro de 2021 Diana Raquel Barroso Gonçalves Rito Virtual reality tool for balance training Dissertação de Mestrado Mestrado Integrado em Engenharia Biomédica Ramo Eletrónica Médica Trabalho realizado sob a orientação de Professora Doutora Cristina P. Santos Doutora Joana Figueiredo iv DIREITOS DE AUTOR E CONDIÇÕES DE UTILIZAÇÃO DO TRABALHO POR TERCEIROS Este é um trabalho académico que pode ser utilizado por terceiros desde que respeitadas as regras e boas práticas internacionalmente aceites, no que concerne aos direitos de autor e direitos conexos. Assim, o presente trabalho pode ser utilizado nos termos previstos na licença abaixo indicada. Caso o utilizador necessite de permissão para poder fazer um uso do trabalho em condições não previstas no licenciamento indicado, deverá contactar o autor, através do RepositóriUM da Universidade do Minho. Licença concedida aos utilizadores deste trabalho https://creativecommons.org/licenses/by-nc-nd/4.0/ v AGRADECIMENTOS E assim termina mais um capítulo da minha vida. O que fez passar de menina a mulher. Foram 5 anos intensos, com muito trabalho, empenho e dedicação depositados no curso que me vai permitir tornar Mestre em Engenharia Biomédica. O percurso nem sempre foi fácil, muitas noites mal dormidas, estudo e trabalhos até às tantas, mas sei que foi por um bom motivo e, olhando agora para trás, tudo valeu a pena. Foram 5 anos onde cresci e me desenvolvi não só em termos profissionais, mas também como pessoa. Onde me dediquei não só ao meu percurso académico, como também ao núcleo de estudantes do meu curso (GAEB), sempre com a esperança e certeza de que poderia dar um pouco de mim, para dar mais um pouco aos outros. No fundo, só queria tornar os 5 anos de alguém tão enriquecedores quanto os meus, repletos de experiências e vivências que vão ficar para sempre. Sim, porque não foram só noites mal dormidas, foram também anos onde me diverti muito e as memórias vão ficar para sempre comigo e para aqueles com quem as partilhei. Restam-me, por isso, os agradecimentos. Não basta só relembrar, mas sim agradecer a quem me proporcionou tudo isto. Os meus maiores agradecimentos vão para a minha família, sobretudo aos meus pais, irmão e avós, que sempre me fizeram acreditar nas minhas capacidades e que festejaram comigo as minhas vitórias. Quero agradecer especialmente à minha mãe, por todo o esforço, empenho e dedicação que sempre teve e continua a ter, comigo e com o meu irmão, para que tenhamos tudo e só nos foquemos nos nossos estudos, com vista a um futuro promissor. Seguidamente, quero agradecer aos meus colegas de curso. Destes, obviamente, com especial destaque para o Miguel, o meu companheiro de todas as horas, no bem e no mal. O meu melhor amigo e namorado. Em seguida, quero agradecer ao Rafael, o meu parceiro de mestrado e amigo, com quem partilhei muitas dores de cabeça em trabalhos sem fim à vista, mas também muitos bons momentos. Deixo ainda menção para o Miguel Castro, uma das melhores pessoas que a universidade me deu, em conjunto com as supramencionadas. Por fim, mas não menos importante, quero agradecer à minha orientadora, professora Doutora Cristina P. Santos, pela orientação e oportunidade de trabalhar neste projeto. Quero também agradecer à Doutora Joana Figueiredo, coorientadora e colega, por toda a ajuda e apoio que sempre me disponibilizou. Por fim, quero agradecer à Cristiana, que foi comigo uma pessoa incansável, super dedicada e que sempre me acompanhou neste último ano. vi STATEMENT OF INTEGRITY I hereby declare having conducted this academic work with integrity. I confirm that I have not used plagiarism or any form of undue use of information or falsification of results along the process leading to its elaboration. I further declare that I have fully acknowledged the Code of Ethical Conduct of the University of Minho. vii RESUMO Em todo o mundo, 15 milhões de pessoas sofrem um acidente vascular cerebral (AVC) por ano. Destas, 66% sobrevivem e metade delas ficam com incapacidade permanente de equilíbrio, limitando a sua independência motora e qualidade de vida. Estes pacientes podem recuperar o seu equilíbrio e independência motora através do fenómeno da neuroplasticidade, alcançado com intervenções de reabilitação. As ferramentas de realidade virtual (RV) podem ser utilizadas como complemento às terapias físicas convencionais, promovendo treinos de alta repetição com estratégias de aprendizagem otimizadas. Assim, os ambientes virtuais podem ser personalizados de acordo com as necessidades iminentes dos pacientes, maximizando a reorganização do cérebro e plasticidade, aumentando a eficácia e acelerando a recuperação do equilíbrio. Não obstante, existe uma falta de ferramentas de RV neste campo, as quais não apresentam uma visão centrada no utilizador. Esta dissertação tem como objetivo conceber, desenvolver e validar uma ferramenta totalmente imersiva baseada em RV, seguindo uma visão centrada no utilizador. A ferramenta de RV desenvolvida inclui quatro desafios virtuais baseados em atividades do dia-a-dia (ADDs), compreendendo um total de nove tarefas motoras. A ferramenta de RV fornece estímulos visuais, sonoros e hápticos, através de óculos de RV, auscultadores incorporados, e comandos vibratórios. Paralelamente, foi realizado um estudo sobre as ADDs mais realizadas e apreciadas, com recurso a um questionário, provando que os desafios virtuais concebidos estão de acordo com a preferência da maioria das pessoas. A partir de uma validação preliminar com sujeitos saudáveis, verificou-se que a ferramenta de RV melhorou significativamente o deslocamento do centro de massa (CDM) na direção mediolateral (ML) e a velocidade mínima do CDM na direção anteroposterior (AP), durante a marcha. Além disso, o deslocamento do CDM e a velocidade máxima e mínima do CDM, nas direções AP e ML, embora não significativas, exibiram melhorias noutras tarefas motoras. Os testes clínicos também revelaram uma melhoria após o treino RV. A avaliação da experiência do utilizador provou a elevada aceitabilidade, valor e utilidade da ferramenta de RV. O trabalho futuro envolve a melhoria da ferramenta com mais e personalizados desafios virtuais e a validação do sistema com doentes durante treinos mais longos. PALAVRAS-CHAVE: ADDs, AVC, CDM, reabilitação de equilíbrio, realidade virtual viii ABSTRACT Worldwide, 15 million people suffer a stroke each year. Of these, 66 % survive and half of them are left with permanent balance disabilities, limiting their motor independence and compromising their quality of life. The patients can recover their balance function and regain their motor independence through neuroplasticity phenomenon, achieved by rehabilitation intervention. Virtual reality (VR) tools may be used as a complement to physical therapies to promote high-repetitive training with optimised learning strategies. Thus, virtual environments can be customised according to the patient’s imminent needs, maximising brain reorganisation, allowing to increase the effectiveness and accelerate balance recovery. Nonetheless, there is a lack of VR tools in this field, and no user-centered design is available. This dissertation aims to design, develop, and validate a fully immersive VR-based tool, following a user-centered design. The developed VR tool includes four virtual challenges based on activities of daily living (ADLs), comprising a total of nine motor tasks. The VR-based tool provides visual, sonorous, and haptic stimuli, through a Head-Mounted Display (HMD), built-in headphones, and vibrotactile controllers. In parallel, a study about the most performed and appreciated ADLs was carried out, using a questionnaire, proving that the developed virtual challenges are in accordance with most peoples’ preferences. From a preliminary validation with healthy subjects, the VR tool significantly improved the user’s center of mass (COM) displacement in the mediolateral (ML) direction, and the minimum COM velocity in the anteroposterior (AP) direction when walking. Moreover, COM displacement and the maximum and minimum COM velocity on both AP and ML directions, although not significantly, also showed improvements in many other tasks. Furthermore, clinical tests revealed improvements after VR training. The user experience evaluation proved the high acceptability, value, and usefulness of the VR-based tool. Future work towards enhancing the VR-based tool with more and customised virtual challenges and extending the VR tool validation with end-users throughout a longer training period. KEYWORDS: ADLs, balance rehabilitation, COM, stroke, virtual reality ix CONTENTS Agradecimentos ................................................................................................................................... v Resumo............................................................................................................................................. vii Abstract............................................................................................................................................ viii List of Figures .................................................................................................................................... xii List of Tables ..................................................................................................................................... xv List of Acronyms ................................................................................................................................ xxi 1. Introduction ................................................................................................................................ 1 1.1 Motivation and Problem Statement ...................................................................................... 1 1.2 Goals .................................................................................................................................. 3 1.3 Research Questions............................................................................................................. 4 1.4 Contribution to Knowledge ................................................................................................... 5 1.5 Dissertation Outline ............................................................................................................. 5 2. Review on virtual reality tools for post-stroke balance rehabilitation ............................................... 7 2.1 Methodology ....................................................................................................................... 7 2.2 Results ................................................................................................................................ 7 2.2.1 VR technology ............................................................................................................ 10 2.2.2 Sensor integration ...................................................................................................... 10 2.2.3 Motor tasks ............................................................................................................... 11 2.2.4 Virtual challenges and control strategies ..................................................................... 12 2.2.5 Clinical outcomes ...................................................................................................... 16 2.3 Discussion ........................................................................................................................ 17 2.4 Conclusions ...................................................................................................................... 18 3. System description: VR-based tool for balance training ............................................................... 20 3.1 VR technology ................................................................................................................... 20 3.2 Sensor integration ............................................................................................................. 22 3.3 Motor tasks ....................................................................................................................... 25 3.4 Virtual challenges and VRE ................................................................................................ 25 3.5 Control strategies .............................................................................................................. 30 3.5.1 First and second person views ................................................................................... 31 xvi Table 6.7 - Mean and SD of T1 and T2, and the p-value of the paired samples test and Wilcoxon test for COM displacement (Disp) and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for collect the food of the kitchen shelves, of the virtual challenge “Cooking”, considering the time condition and a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold ................................................................................................................................................. 71 Table 6.8 - Number of ingredients caught in kitchen’s shelves in each trial of the virtual challenge “Cooking” ................................................................................................................................... 71 Table 6.9 - Mean and SD of T1 and T2,, and the p-value of the paired samples test and Wilcoxon test for COM displacement (Disp) and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for stand-to-sit, sit-to-stand, and walking motor tasks of the virtual challenge “Watch Tv”, considering the time condition and a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold ......................................................................................................................... 74 Table 6.10 – Mean and SD of the ROM of the left and right knee joint angles on the sagittal plane (degrees) for stand-to-sit, sit-to-stand, and walking motor tasks, and the stride length (cm) and stride time (s) for walking motor task in each trial of the virtual challenge “Watch Tv” .................................................. 75 Table 6.11 – Television state achieved in each trial of the virtual challenge “Watch Tv” ..................... 76 Table 6.12 - Mean and SD of pre and post VR training trials, and the p-value of the paired samples test and Wilcoxon test for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for BBS’ tasks, considering the time condition and a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold ................................................. 78 Table 6.13 - Mean and SD of T1 and T2, and the p-value of the paired samples test and Wilcoxon test for COM displacement (Disp) and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for sit-to-stand, walking, and stand-to-sit motor tasks of TUG, considering the time condition and a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold 83 Table 6.14 - Mean and SD of the ROM of the left and right knee joint angles on the sagittal plane (degrees) for sit-to-stand, stand-to-sit, place alternate foot on step, and standing on one leg, and the maximum hand position (cm) for reaching forward BBS’s tasks ............................................................................... 84 Table 6.15 - Mean and SD of the ROM of the left and right knee joint angles on the sagittal plane (degrees) for sit-to-stand, walking and stand-to-sit motor tasks, and the stride length (cm) and stride time (s) for walking motor task in each trial of TUG clinical scale ......................................................................... 85 Table 6.16 – Final scores for BBS and TUG clinical tests, for each of the participants, before and after the VR intervention ........................................................................................................................... 85 xvii Table 6.17 - Mean and SD for pre and post VR training, the p-value of the paired samples test, and the p-value of the Wilcoxon test for BBS and TUG clinical tests, considering the time condition and a significance level of 5 % .................................................................................................................... 86 Table 6.18 - Mean and standard deviation (SD) of all the participants, for each factor of IPQ............... 89 Table 6.19 - Mean and standard deviation (SD) of all the participants, for each subscale of IMI 90 Table 6.20 - Participants’ answers to the open questions of the Value/Usefulness subscale of IMI questionnaire ................................................................................................................................... 90 Table 0.1 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), ROM of both left and right knee joint angle on the sagittal plane (degrees), and number of steps climbed by each participant, in both trials, while climbing stairs (CS), in Fruit Catcher virtual challenge ............................................................................................................................. 104 Table 0.2 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), maximum hand position (m) for left, forward and right directions, and number of apples caught by each participant, in both trials, while catching fruits (CF), in Fruit Catcher virtual challenge ............................................................................................................................. 105 Table 0.3 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), ROM of both left and right knee joint angle on the sagittal plane (degrees), and number of times each participant kicked the bathtub rim, in both trials, while entering to the bathtub (EB), in Take a Shower virtual challenge .......................................................................................... 106 Table 0.4 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), ROM of both left and right knee joint angle on the sagittal plane (degrees), and the amount of time the sound played (s) 10 s being the total, in both trials, while each participant is inside the bathtub (IB), in Take a Shower virtual challenge ............................................................... 107 Table 0.5 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), and the number of ingredients caught, in both trials, by each participant, in Cooking virtual challenge............................................................................................................. 108 Table 0.6 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), and ROM of both left and right knee joint angle on the sagittal plane (degrees) of each participant while sitting, in both trials, in Watch Tv virtual challenge ...................... 109 Table 0.7 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), and ROM of both left and right knee joint angle on the sagittal plane (degrees) of each participant while standing, in both trials, in Watch Tv virtual challenge .................. 110 xviii Table 0.8 – Mean and standard deviation of COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), ROM of both left and right knee joint angle on the sagittal plane (degrees), stride time (s), and stride length (m) of each participant while walking, in both trials, and if he turn on the television, in Watch Tv virtual challenge .......................................... 111 Table 0.9 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), ROM of both left and right knee joint angle on the sagittal plane (degrees), and the BBS score of each participant, for BBS tasks 1,4,12, and 14, respectively, before and after VR training ........................................................................................................................................... 113 Table 0.10 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), maximum hand forward position (cm), and the BBS score of each participant, for BBS task 8, before and after VR training ................................................................... 115 Table 0.11 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), and the BBS score of each participant, for the fourteen BBS tasks 2, 3, 5, 6, 7, 9, 10, 11, 13, respectively, before and after VR training ...................................................... 116 Table 0.12 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), and ROM of both left and right knee joint angle on the sagittal plane (degrees) of each participant while performing the standing TUG task, in both trials, before and after VR training ........................................................................................................................................... 121 Table 0.13 - Mean and standard deviation of COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), ROM of both left and right knee joint angle on the sagittal plane (degrees), stride time (s), and stride length (m) of each participant while performing the TUG walking task, before and after VR training ........................................................................... 122 Table 0.14 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), and ROM of both left and right knee joint angle on the sagittal plane (degrees) of each participant while performing the TUG sitting task, before and after VR training 124 Table 0.15 - The p-values of the normality tests for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for CS, and CF motor tasks of the virtual challenge “Fruit Catcher”, considering a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold .................................................................................................................... 125 Table 0.16 - The p-values of the normality tests for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for EB, and standing on one leg inside the xix bathtub (IB) motor tasks of the virtual challenge “Take a Shower”, considering a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold ...................................................... 126 Table 0.17 - The p-values of the normality tests for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, collecting food from the kitchen shelves of the virtual challenge “Cooking”, considering a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold ....................................................................................................... 127 Table 0.18 - The p-values of the normality tests for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for stand-to-sit, sit-to-stand, and walking motor tasks of the virtual challenge “Watch Tv”, considering a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold ...................................................................... 128 Table 0.19 - The p-values of the normality tests for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for BBS motor tasks, considering a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold ................. 129 Table 0.20 - The p-values of the normality tests for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for sit-to-stand, stand-to-sit, and walking motor tasks of TUG, considering a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold ....................................................................................................................... 134 Table 0.21 - The p-values of the normality tests for BBS and TUG scores, considering a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold ........................................... 135 Table 0.22 - The p-values of the homogeneity test for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for CS, and CF motor tasks of the virtual challenge “Fruit Catcher”, considering a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold .................................................................................................................... 135 Table 0.23 - The p-values of the homogeneity tests for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for EB, and standing on one leg inside the bathtub (IB) motor tasks of the virtual challenge “Take a Shower”, considering a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold ............................................... 136 Table 0.24 - The p-values of the homogeneity tests for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, collecting food from the kitchen shelves of the virtual challenge “Cooking”, considering a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold ....................................................................................................... 137 xx Table 0.25 - The p-values of the homogeneity tests for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for stand-to-sit, sit-to-stand, and walking motor tasks of the virtual challenge “Watch Tv”, considering a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold ..................................................................... 138 Table 0.26 - The p-values of the homogeneity tests for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for BBS motor tasks, considering a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold ................. 139 Table 0.27 - The p-values of the homogeneity tests for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for sit-to-stand, stand-to-sit, and walking motor tasks of TUG, considering a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold ....................................................................................................................... 143 Table 0.28 - The p-value of the homogeneity test for TUG scores, considering a significance level of 5 %. The most representative results (p-value < 0.05) appear in bold ...................................................... 143 xxi LIST OF ACRONYMS 10 mWT – 10-meter Walking Test 3D – Three Dimensional 6 mWT – 6-minute Walking Test ABC – Activities-specific Balance Confidence scale (ABC) ADLs – Activities of Daily Living AP – Anteroposterior BBA – Brunel Balance Assessment BBS – Berg Balance Scale BI – Barthel Index BiRD Lab – Biomedical Robotic Devices Laboratory BPM – Balance Performance Monitor BOS – Base of Support CB&M – Community Balance and Mobility Scale CMEMS – Centre of MicroElectroMechanical Systems CMOS – Complementary Metal-Oxide Semiconductor COM – Center of Mass COP – Center of Pressure CTSIB – Clinical Test for Sensory Interaction in Balance DGI – Dynamic Gait Index EMG - Electromyography FAC – Functional Ambulatory Category FAI – Frenchay Activity Index FIM – Functional Independence Measure FMA-LE – Fugl-Meyer Assessment oriented to Lower Extremity FRT – Functional Reach Test FSST – Four Step Square Test FTSST – Five Times Sit-to-Stand Test HMD – Head Mounted Display IMI – Intrinsic Motivation Inventory IPQ – Igroup Presence Questionnaire IR - Infrared xxii K-MBI – Korean version of Modified Barthel Index LRT-L/R – Lateral Reach Test Left/Right MBI – Modified Barthel Index ML – Mediolateral MMAS – Modified Motor Assessment Scale MMT – Manual Muscle Test MoCA – Montreal Cognitive Assessment POMA – Tinetti Performance-Oriented Mobility Assessment PPT – Physical Performance Test RGB - Red-Blue-Green RNLI – Reintegration to Normal Living Index (RNLI) ROM – Range of Motion RT – Romberg’s Test SIS – Stroke Impact Scale sRT – sharpened Romberg’s Test STOLL – Standing on The Left Leg STORL – Standing on The Right Leg TUG – Timed Up and Go Test VR – Virtual Reality VRE – Virtual Reality Environment WAQ – Walking Ability Questionnaire 1 1. INTRODUCTION This dissertation presents the work developed in the scope of the fifth year of the Integrated Master’s in Biomedical Engineering, at the University of Minho, during the academic year of 2020-2021. The academic year was passed working in the Biomedical Robotic Devices Laboratory (BiRD Lab), included in the Centre of MicroElectroMechanical Systems (CMEMS) Research Centre, at the University of Minho, Guimarães, Portugal. During this period, it was developed and validated a fully immersive virtual reality (VR) tool for balance training. This system was developed to improve the static and dynamic balance aiming at the rehabilitation process’s acceleration. All the methods, results, and conclusions are detailed in this document. 1.1 Motivation and Problem Statement Stroke affects annually 15 million people worldwide, predominantly over 40 years old and in 8% of children with sick cell disease [1]. Of these 15 million, 5 million survive but are left permanently disabled [1]. Due to this, stroke is the leading cause of serious long-term disability, usually including loss of balance, postural instability, muscle paralysis or muscles weakness, and impaired walking ability [2]. This and other neurological conditions (i.e., Parkinson’s disease, cerebral palsy), ear disorders (i.e., ear infections, vestibular problems as inner ear abnormalities or Meniere’s disease), head injury, and age (>= 60 years old), as well as certain medications are the main leading causes of loss of balance [3], [4]. Balance is the ability to remain in a position (e.g., standing or sitting) or to move without losing postural control or falling [5]. Good balance requires the coordination of several parts of the body, such as muscles, bones, joints, central nervous system, and inner ear [6], which are usually affected by the problems mentioned above. Balance disorders are one of the main risk factors for falling, being the second leading cause of unintentional injury deaths worldwide [7], mostly in people over 65 years old [8]. The loss of balance itself and the associated fear of falling lead to people’s loss of independence and increased difficulty to perform their activities of daily living (ADLs), compromising the quality of life, and professional and social inclusion [9]. According to the current situation, there is a need to focus on the improvement of balance recovery strategies. People who survived and were left with disabilities can recover their balance function and regain their motor independence through neuroplasticity phenomenon. Neuroplasticity is the ability of the brain to redevelop its neural network and rewire functions at the healthy part of the brain by forming new 2 connections and pathways [10] to take over the role of the damaged part [11]. However, for this phenomenon to occur, it is necessary to stimulate the cells of the brain through exposure to frequent and repetitive activities [11], [12]. This can be achieved by rehabilitation intervention, either physical or robotic [11], [13]. Physical therapy, as conventional balance therapy, involves changing positions, breathing exercises, and exercise therapy in passive and active mobilization, with the guidance of a therapist [11]. Conventional rehabilitation has the disadvantages of being non-standard and dependent on the therapist’s preference [11], [13]. Robotic therapy includes robotic devices with advanced control techniques, which allow interaction with the user, provide body-weight support, intense high-repetition training over a longer period, and optimised learning strategies, in opposition to conventional therapies [14]. Furthermore, some robotic devices are equipped with embedded sensors providing objective information about the user’s motor condition and evolution of the user’s performance, and also giving feedback to the user [14]. Feedback about the user’s performance during training not only increase her/his motivation, but also facilitate the neuroplasticity phenomenon [15], and may improve the therapy’s efficacy [16]. Moreover, robotic therapy is less therapist-dependent once a single physiotherapist can properly supervise multiple patients who train on different robotic devices [14]. However, the physiotherapists perform an essential and irreplaceable work once training with clinicians supervising show more clinically significant improvements than those without supervision [16]. There is evidence that combining both robotic and physical therapies improves and accelerates the user’s recovery, once robot-assisted rehabilitation provides a standardized environment for all the patients, where therapy intensity and level of difficulty are adjusted according to the user’s imminent needs [14], [15]. Current scientific works demonstrate that virtual reality (VR) can be a promising and powerful tool, as a robotic rehabilitation tool, in order to find an optimal solution for the rehabilitation of cognitive and motor functions [17]–[19]. VR offers the possibility to create endless three-dimensional (3D) virtual reality environments (VREs), as close as desired to the real ones, eliciting realistic perceptions and reactions in the user [17], [19]. Moreover, it is possible to design high-detailed environments that can be too dangerous, expensive, or impossible to create in physical reality by under full control and safety conditions to patient and the therapist [17]. Furthermore, VREs can be designed and customised by designing virtual challenges according to the user’s imminent needs, which is critical for maximising brain reorganization and neuroplasticity phenomenon [20]. 3 The user can experience more or less immersion in the VRE when using a Head-Mounted Display (HMD) or single screen, respectively [17]. Fully immersive scenarios bring the user inside the VRE, allowing her/him to feel that she/he is physically present in the observed environment [17]. This characteristic, also called “Sense of Presence”, can be useful once it allows obtaining realist physiological reactions from the user to virtual stimuli, as if the subject was really in a place identical to the virtual one [17]. On the other hand, fully immersive scenarios allow a more natural interaction with the surrounding environment [17]. Additional biosensors such as Inertial Measurements Units (IMUs), force sensors, and electromyography (EMG) sensors, by tracking the full-body patients’ kinematics, movement dynamics, and muscle activation [20], respectively, allows the user to interact with the virtual environment using her/his entire body as an active part of the 3D virtual world [17]. VR intervention has already been demonstrated to be effective in improving balance, physical disabilities, functional ability, and muscle strength in a large number of studies [21]. Moreover, some significant improvements were also obtained in clinical outcomes, such as Berg Balance Scale and Timed Up and Go Test, after VR intervention on post-stroke patients [21]. Furthermore, the satisfaction, adherence, engagement, enjoyment, and low pain of VR therapies can be highlighted, contrarily to the physical therapies [21]. However, these systems are designed for entertainment purposes, being this its first and main objective. Thus, there is a lack of immersive VR systems in the balance rehabilitation field, and there is yet no VR tool with a user-centered design that enables balance training oriented to patients’ needs in ADLs. Further, most of studies used non-wearable VR technology and biosensors, involved to provide real-time feedback of patients’ movements, limiting the VR tools to clinical practice to an indoor fixed facility. In this manner, there is a need to focus on the development of a wearable VR-based tool with a user-centered design for balance rehabilitation, aiming the acceleration of the recovery process. 1.2 Goals The ultimate goal of this dissertation is to design, develop, and validate a fully immersive and wearable VR-based tool to be used as a promising complementary tool of physical balance therapy, aiming at the acceleration of the balance rehabilitation process. Fully immersive VREs will be developed to be the most realistic and identical as possible to the real world to obtain realist physiological reactions from the user to virtual stimulus. The user will be immersed through VR glasses in VREs controlled in real-time by inertial data from wearable sensors, providing feedback and, consequently, encouraging the users to execute static and/or dynamic balance tasks to complete the virtual challenges induced by an enthusiast serious game. User-centered strategies will be used to design the VREs according to the users’ imminent 10 2.2.1 VR technology The VR technology used in the reviewed studies provides visual (28 studies) and/or auditory (21 studies) cues to patients. The study [39] is the only which provides vibrotactile cues through a pager vibrator. By analysing Table 2.1, it is possible to visualize three types of display systems (i.e., screen, a combination of screens, HMD) commonly used to provide visual feedback to patients in VR-based balance rehabilitation. Twenty two studies provided access to the virtual environment through a television (TV) screen [24], [25], [30], [36]–[38], [43], [48], [50], LCD [29], [45], [47], non-discriminated screen [26]– [28], [31], computer monitor [42], non-discriminated monitor [37], [40], [49], projector [51], and video display [33]. Four studies used HMD [34], [39]–[41], one of which discriminated it as being the Virtual Research V6 HMD [39]. Two studies propose a combination of three connected screens [44] or projectors [49]. Thirteen studies provided auditory cues through TV screen [24], [25], [36], [38], [43], [48], [50], LCD [45], [47], computer monitor [42], and non-discriminated monitor [32], [35], [46], all with built-in speakers. Five studies used external sound devices, such as loudspeakers [26], [27], speakers [33], 3D auditory outputs [44], and a sound system [49]. Three studies used HMD with built-in headphones [39]– [41]. The remaining seven studies do not report auditory cues [28]–[31], [34], [37], [51]. 2.2.2 Sensor integration Most studies propose VR-tools with integrated sensors that allow patients to interact with the virtual environment and receive real-time feedback of their movements. Four studies do not mention sensor integration [26], [27], [40], [41]. Fifteen studies used a camera and four of them with reflective markers. One study utilized two OptiTrack FLEX:C120 cameras (NaturalPoint, OR) belonging to the BioTrak VR system to estimate the 3D position of two markers fixed to the participant’s insteps [33]. Eight studies used an infrared camera: one from Balance Control Trainer (BCT) system to measure the four markers’ position on the knee [31], and six studies used the Kinect sensor (also works as Red-Green-Blue [RGB] camera) into the Xbox system to monitor the patient’s COM (unique position where the sum of the weighted position vectors of all the parts of a system is equal to zero) [31], [32], [41], [43], [46], [47]. Mazzini et al. [49] used six infrared cameras from the Stability and Balance Learning Environment system (Motekforce Medical, Amsterdam, Netherlands). Two studies utilized a camera from the IREX system (Vivid group, Toronto, Canada), which determines the markers’ position of cyber gloves [30], [37]. One study used a web video camera from 11 the SeeMe system (Brontes Processing, Gliwice, Poland) that captures the user’s movements [48]. Xu et al. [46] involved the PrimeSense 3D Awareness Sensor (Apple Inc, USA) that provides gesture and skeleton tracking. Two studies employed a non-discriminated camera [29], [39] that records the leg and foot movements. Jaffe et al. [39] also used flat foot switches to detect collisions. Seven studies used a balance board that measures, in real-time, the patient’s center of pressure (COP) (point of application of the ground reaction force vector). Six of the boards belong to the Nintendo Wii Fit [3], [10], [13], [22], [27], [31] and the other is included in the BalPro system (Man&Tel, Gumi, Korea) [28]. This last study also utilized a tilting sensor that estimates the knee joint angle [28]. Mazzini et al. [49] used a force platform, included in the Stability and Balance Learning Environment system, with the purpose of measuring the patient’s COP. Lee et al. [31] used two electronic scales included in BCT system. In another study, these authors [32] used a platform belonging to the BioRescue system (RM Ingénierie, Rodez, France) to monitor the patient’s COP. One study employed the Tetra-ataxiometric posturography (Tetrax) (Sunlight Medical Ltd., Ramat Gan, Israel) that measures postural sway using the change in weight burden onto each of four force plates included in the system [37]. One study used the Stewart platform included in the Rutgers Ankle Rehabilitation System to determine the ankle angle [42]. Yang et al. [44] used an electromagnetic system (Fastrack, Polhemus), which tracks leg motion. Park et al. [34] used a non-discriminated sensor that measures the patient’s movement. 2.2.3 Motor tasks Looking at Table 2.1, it is possible to verify that VR-based tools comprise different motor tasks for balance rehabilitation, namely: weight shifting/bearing (9 studies), walking (7 studies), stepping (6 studies), specific-body movement at pelvic/hip (6 studies: 4 standing [1 with a harness], 1 sitting, and 1 sitting and in supine position), knee (6 studies: 5 standing [3 with and 2 without harness] and 1 not mentioned), feet (4 studies: 2 sitting, 1 standing with a harness, and 1 not mentioned), trunk (3 studies: 2 standing [1 with a harness], 1 sitting, and 1 not mentioned), arms/hands (2 studies), and standing on a single leg (2 studies). Non-discriminated aerobic exercises, reaching exercise, upper extremity movement, sit-to-stand, squats, dodging left and right, ducking, and jumping are referred in single studies. Fifteen studies combined more than one motor task [28]–[32], [34], [35], [39], [42], [45]–[47], [49]– [51] and four studies did not mention any motor task [24], [25], [38], [43]. Seven studies addressed walking, all of them on the treadmill [26], [27], [39]–[41], [44], [48]. Five of them applied different speeds according to the patient’s evolution [26], [27], [40], [41], [44], and only one study varied the slope in conjunction with the virtual environment [44]. Six studies addressed stepping 12 [30], [33], [39], [47], [49], [50]. Nine studies addressed weight shifting in the horizontal direction [28], [31], [32], vertical direction [31], [32], and non-discriminated direction [30], [35], [36], [47], [49], [51]. Two studies tackled weight-bearing training [34], [35]. Rajaratnam et al. [36] issued COM change in sitting and standing positions. Six studies addressed pelvic/hip movement [29], [30], [34], [35], [46], [50] including pelvic tilt in supine and sitting position [34], hip abduction [29], [35], [50], adduction [29], flexion [35], [46], extension [46], and external-internal rotation [35]. Six studies addressed knee movement [28], [30], [31], [35], [46], [49] focused on knee flexion [28], [31], [35], [46], [49] and extension [31], [35], [46], [49]. Three studies tackled trunk movement [30], [34], [35]. Park et al. [34] mentioned maintenance of trunk stability in supine and standing positions and trunk upright control in sitting position. In another study, these authors exploited trunk rotation [35]. Four studies addressed feet movement [29], [30], [35], [42], tackling dorsiflexion and plantarflexion [29], [35], [42], adduction and abduction [29], and inversion and eversion [42]. Two studies addressed arms/hands movements [49], [50], including arm circles [50]. Park et al. [34] also addressed lower extremity muscle strengthening exercises and upper extremity active movement. IREX VR system includes five tasks for moving COP [37]. Tetrax system enables left-right and anterior-posterior weight shifting and weight-bearing [37], [52]. Nintendo Wii Fit system, in [24], [25], [38], [43], [51] includes balancing on one leg, leaning, rotating, and moving body to a rhythm [53]. 2.2.4 Virtual challenges and control strategies Virtual challenges are intended as tasks taking place in virtual environments controlled to work as serious games. In [26], [27], post-stroke patients walked on a treadmill while an unsynchronized real-world video recording plays, depicting a sunny or rainy 400-m walking track, a 400-m walking track with obstacles, daytime or night-time walks in a community, or walking on trails. Similarly, VR-based tools from [40] and [41] simulate a park scroll and street walking, respectively. However, in [41], the optic flow speed is configured according to the 10-m walking test. Yang et al. [44] presented a virtual environment simulating the following scenarios from a typical community in Taipei: lane walking, street crossing, obstacles striding across, and park scroll. The patients are encouraged to uphill and downhill walking (treadmill slope was enabled), fast walking, and step over obstacles while walking on an unsynchronized treadmill. The patients’ leg motions were tracked during the obstacles striding across the scenario for providing auditory feedback when detecting collisions with the virtual obstacles. In all the above studies, if the patient was 13 able to walk stably for more than 20 seconds, the treadmill speed was increased by 0.1 km/h, every 20 seconds [40], [41] and 5 % during the subsequent training session [26], [27], [44]. In [39], patients were instructed to step over ten identical stationary virtual obstacles while walking on the treadmill. They received real-time visual feedback of their legs’ lateral view and vibrotactile and short sound auditory feedback when a collision with the virtual obstacle occurs (measured when the foot should not be on the ground). The virtual environment from [33] represents the patients’ feet with two shoes mimicking their real movement through the positions of the markers placed on each foot. The patients are encouraged to stepping by reaching with one foot the items that rose from the ground while maintaining the other foot within a circle. In [48], the patients walked at the same speed while training with three VR games: Ball Game, Reactive Boxing, and Cleaning Windows, included in the SeeMe system. In the Ball Game, patients are instructed to strike virtual balls with their upper extremity and avoid possible virtual shoes that can appear randomly from different directions. In Reactive Boxing, the users must touch the virtual boxes, that appear randomly on both sides of the screen, within a specific period. In Cleaning Windows game, the participants were required to clean a series of windows, as quickly as possible, by wiping off the virtual dirt that covered the windows. The users’ movements were tracked by a web video camera, allowing the interaction with the VRE. Study [29] asked for the patients to observe the real-time movements of the unaffected limb on the monitor and mimic these movements with the affected limb while sitting on a mat without back support. Similarly, in [34], the patients observe in real-time their posture and movement, and they are encouraged to mimic a pre-recorded reference motion. In [28], the patient’s COP and knee flexion are traduced into a horizontal and vertical hand-shaped cursor movement on a screen, respectively. The patients are encouraged to perform weight shifting and knee flexion/extension by moving the screen's cursor to catch fruits (Fruit-Harvesting game). Similarly, in [31], the patients move horizontally and vertically a virtual eraser according to the measurements of electronic scales and position of the markers placed on the knee, respectively. The virtual challenge consists of achieving the maximum score, defined as the percentage of the board cleaned in 2 min (Board Cleaner Game). In [42], the patients are encouraged to perform ankle movements while sitting to navigate a virtual plane or a boat towards a series of targets according to ankle joint angle. Lee et al. [32] controlled the virtual environments using patients’ COP and encourage patients’ weight shifting during the following games: City Walking (left-right shifting), Hot Air Balloon (up-down 14 shifting), and Bubble (total shifting). The IREX VR system, in [30], comprises the following games: Stepping up/down, Sharkbait, and Snowboard, which encourages patients’ weight shifting, stepping, trunk, pelvis/hip, knee, and ankle movements. In [37], the IREX VR games are not specified. In both studies, patients are reflected on the screen, interacting with the virtual environment according to the markers' position from the glove and patients’ COP. In [37], Tetrax games are also not specified, but according to another study [52], the Tetrax system comprises the following games: Catch, Skyball, Tag, Gotcha, Speedball, Immobilizer, Target, and Freeze, which encourage the patients to move accordingly to the COP. The Xbox Kinect system, used in [35], [36], [45] includes the following games: Boxing, Table Tennis, Soccer, Golf, Ski, Football, 20 000 Leaks, River Rush, and Reflex Ridge, which recognize patients’ COM. During Boxing, the patients are encouraged to punch virtual objects by moving their arms. In Table Tennis, Soccer, and Golf, the patients must hit a virtual ball with a virtual racket, kick a virtual ball, and put virtual balls in virtual golf holes, respectively. During Ski, the users must avoid the virtual barriers and follow the slope by shifting their weight. In Football game, the patients are encouraged to run with a virtual ball, avoiding the virtual opponent players. In 20 000 Leaks game, the patients must use their body’s movement to stop leaking water as it spills through the tank walls in the game. In the two-players game River Rush, the patients must control the raft, while rushing down a river by moving their body from side to side or jumping to avoid obstacles. In Reflex Ridge, the patients are also encouraged to avoid obstacles by dodging left and right, ducking, or jumping over hazards as they spring forward. The Nintendo Wii system, in [24], [25], [36], [38], [43], [51], comprises the following games: Soccer Heading, Ski Slalom, Balance Bubble, Hula Hoop, Ski Jump, Table Tiling, Penguin Slide (included in the Nintendo Wii Fit, using a balance board that recognizes patients’ COP) [24], [25], [36], [38], [43], Boxing, Bowling, Snowboarding, Swimming, and Tennis (included in the Nintendo Wii Sports, using a Kinect sensor that recognizes patients’ body movement) [51]. Soccer Heading game encourages the patients to reach virtual balls flying at them and to avoid other flying objects such as cleats and panda heads. In Balance Bubble game, the patients try to avoid virtual obstacles such as walls, rocks, and bees, while going down a virtual river. During Hula Hoop, Penguin Slide, Table Tilting, Boxing, Bowling, and Tennis, the players must catch virtual bows, virtual fish that comes off the water, tilt virtual balls into holes, hit the opponent towards knocking her/him on the ground, knock down virtual pins with a virtual ball, and hit a virtual ball with a virtual racket, respectively. When performing Ski Slalom (or Snowboarding) and Ski Jump, the patients are encouraged to navigate following virtual flags and to jump off a virtual hill, respectively. Swimming game encourages the patients to simulate swimming exercises with Wii remotes. 15 Xu et al. [46] developed a depth camera-based task-specific game called Stomp Joy, where patients are encouraged to stand on a single leg for stepping on gophers, performing hip and knee flexion or extension to control the size and falling of a footprint, respectively. The footprint changes to pink from white, and a gong sound plays if the joint angles meet the requirements. In the opposite case, a mocking sound is reproduced. Cikajlo et al. [47] designed and developed a Rehabilitation Wayout in Responsive Home Environments – REWIRE system, consisting on three VR games: Animal Hurdler, Fruit Catcher, and Horse Runner, recognizing patients’ COP and COM through the Wii Fit balance board and the Kinect sensor, respectively. In Animal Hurdler game, the patients are encouraged to step over small creatures approaching by raising a foot (standing on a single leg). In Fruit Catcher game, the patients must catch fruits and avoid chocolate eggs falling from the trees by performing weight shifting and stepping. In the Horse Runner, the patients must drive a horse that runs in the woods avoiding hitting the branches with the avatar’s head by performing squats. Standing up makes the horse to run faster and the user receives bonus scores as floating honey jars. In [49], Mazzini et al. proposed the Stability and Balance Learning Environment composed of eight VR games. The games were divided into two blocks of sessions: odd sessions that include Balloon Pop, City Ride, Hit the Mole, and 2D Maze, and even sessions that include Road Encounters, Road Stepping, Paper Flight, and Hit Knees. In the Balloon Pop and Hit Knees, the patients must burst the largest number of balloons and hit spheres that appear in a semi-circle on the screen, respectively, by performing hands’ movements and, only in Hit Knees game, knee flexion and extension. In Road Encounters and Road Stepping, the patients are encouraged to hit the largest number of birds that appear on the screen by moving their hands, while automatically transverses a road, and while move the avatar using the frequency of movements of stationary march, respectively. In City Ride, 2D Maze, and Paper Flight, the patients must drive a car in a city avoiding collisions with other vehicles, drive a red ball through a labyrinth aiming to complete the course and avoiding collisions with the walls, and drive a paper airplane through a tunnel to pass as many as possible inside the narrower rings, respectively. In Hit the Mole game, the patients are encouraged to perform stepping on the force platform to hit with a hummer the moles that arise from one of the four burrows. The interaction between the users and the VRE was provided by a force platform and infrared cameras, measuring the patients’ COP and the position of reflective markers placed on the patients’ hands. Sheehy et al. [50] used a Kinect camera that captures the participants’ movements allowing to control an avatar. This study includes several games and activities to train subjects in standing balance 16 (e.g., moving a ball along maze, and Slalom Skiing), reaching (e.g., putting dishes away), stepping (e.g., stepping onto tablets placed in a circle, and Whack-a-Mole), gentle strengthening (e.g., standing hip abduction, arm circles, and sit-to-stand), and aerobic exercises (e.g., marching on the spot). 2.2.5 Clinical outcomes The clinical outcomes are performed by a physiotherapist to record a clinical profile of the patient before or after a rehabilitation session to follow the evolution of patients' performance during the recovery process. Table 2.1 shows that some clinical outcomes are addressed in multiple studies, namely: Berg Balance Scale (BBS: static and dynamic balance excluding walking) (22 studies [20]–[29], [31]–[34], [38], [39], [41], [42], [44]–[47]), Timed Up and Go Test (TUG: dynamic balance including walking) (18 studies [20]–[25], [27], [28], [31], [32], [34], [36], [37], [41], [43]–[46]), 10-meter Walking Test (10 mWT: functional mobility and vestibular function) (11 studies [25]–[27], [29]–[31], [37]–[39], [43]–[45]), 6-minute Walking Test (6 mWT: aerobic capacity and endurance) (5 studies [28], [39], [41], [42], [49]), Functional Reach Test (FRT: dynamic balance during maximal forward reach) (5 studies [29], [36], [38], [41], [48]), Activities-specific Balance scale (ABC: confidence during ambulation) (4 studies [38], [40], [44], [48]), Functional Ambulation Category (FAC: walking on stairs and level ground) (3 studies [28], [31], [43]), Lower Extremity Fugl-Meyer Assessment (FMA-LE: motor and sensory functions, balance, joint range of motion and pain) (3 studies [35], [42], [46]), Functional Independence Measure (FIM: independence for self-care, transfers, locomotion, communication, and social cognition) (2 studies [24], [51]), Modified Barthel Index (MBI: functional independence during ADLs) (2 studies [36], [46]), Korean version of Modified Barthel Index (K-MBI) (2 studies [28], [31]), Manual Muscle Test (MMT: muscle strength during knee extension) (2 studies [28], [31]), Tinetti Performance-Oriented Mobility Assessment (POMA: balance and stability during ADLs) (2 studies [33], [39]), and Stroke Impact Scale (SIS: disability and health-related quality of life) (2 studies [49], [50]). Single studies also referred Barthel Index (BI) [43], Balance Performance Monitor (BPM: static and dynamic balance) [30], Modified Motor Assessment Scale (MMAS) [30], Brunel Balance Assessment (BBA) [33], Dynamic Gait Index (DGI: ability to control balance while walking in the presence of external demands) [38], Frenchay Activity Index (FAI: instrumental ADLs that patients have undertaken in the recent past) [38], Physical Performance Test (PPT: physical function while performing ADLs) [39], 7meter walkway [42], Four Step Square Test (FSST: stepping over objects sideways, forward, and backward) [47], Standing on the left leg (STOLL) [47], Standing on the right leg (STORL) [47], Romberg’s 17 Test (RT: vision, proprioception, and vestibular sensory function while balancing) [47], sharpened Romberg's Test (sRT) [47], Lateral Reach Test Left/Right (LRT-L/R: similar to FRT but with maximal lateral reach) [48], Clinical Test for Sensory Interaction in Balance (CTSIB: postural control) [47], Montreal Cognitive Assessment (MoCA: attention, concentration, executive functions, memory, language, constructional visual skills, conceptual thinking, calculations, and logical reasoning) [49], Five Times Sitto-Stand Test (FTSST) [50], Community Balance and Mobility Scale (CB&M) [50], Reintegration to Normal Living Index (RNLI: degree of reintegration back into normal social activities after illness) [50], and Walking Ability Questionnaire (WAQ: mobility at home and community) [44]. 2.3 Discussion Most of the reviewed studies used a single screen. This preference can be explained once a screen is generally a cost-effective solution for a VR-based tool when compared to multiple screens and even more an HMD. Moreover, in opposition to an HMD, the screen allows the physiotherapists to follow in real-time the patients’ performance during the balance training and, consequently, efficiently provide additional support to patients [54]. However, the combination of screens and HMD provides greater immersion in the virtual environment than the screens. The full immersion in the virtual environment promotes patients’ concentration, involvement, and active participation in balance training, leading to an efficient recovery [55]. Furthermore, in opposition to the screens, the HMD is a wearable setup device. Despite the promising potentialities of HMD, only four studies have involved this technology. Moreover, some studies described the use of auditory (mainly) [3], [8]–[10], [13], [14], [18], [19], [21], [22], [24]– [34] and vibrotactile cues [39], filling the gap of auditory and haptic interaction, respectively, with the virtual environment. Most of the VR-based tools include sensors to provide real-time feedback of patients’ movements and to allow the patients to effectively interact with the virtual environment. Those who did not mention the use of sensors [26], [27], [40], [41] only use VR technology to develop unsynchronized virtual environments. Most of the reviewed studies used a camera or a balance board/platform as sensors. They have a promising impact in balance training once they provide an objective assessment of patients’ position, posture, COM, and movement in the case of a camera and patients’ COP in the case of a balance board/platform. However, both cameras and balance boards/platforms are non-wearable, limiting the VR tools to clinical practice to an indoor fixed facility. VR-based tools were exploited to guide the execution of different motor tasks, including walking, stepping, weight shifting/bearing, pelvic/hip, knee, feet, trunk, arms/hands movement, and standing on 18 a single leg. Weight shifting/bearing, treadmill walking, stepping, pelvic/hip and knee movement were the most performed motor tasks, highlighting the role of both upper and lower limbs in balance control. Most studies [28]–[32], [34], [35], [39], [42], [45]–[47], [49]–[51] combined at least two motor tasks, enabling a more holistic balance training, personalized according to the patients’ imminent needs and indispensable for an independent daily living [16]. Treadmill walking was mainly addressed alone in opposition to the remaining motor tasks. Moreover, the VR-based tools that encourage treadmill walking use sensors mainly to develop a virtual environment. Thus, there is space to develop VR-based tools combining walking with other motor tasks and integrating sensors to provide real-time feedback and allow patients’ interaction with the virtual environment. Most of the VR-based tools comprise a closed-loop control, providing real-time feedback according to integrated sensors' objective measures. Feedback encourages the patients to self-control their movements towards motor relearning [56]. The virtual challenges applied in closed-loop VR-based tools have no trend but are dependent on the encouraged motor task. These challenges imply the development of serious games generally aiming to catch, avoid, or navigate virtual objects, fostering patients’ enthusiasm and motivation. Thus, patients are encouraged to perform implicit motor tasks to accomplish the virtual challenge, improving their balance control. Although the variety of VR-based tools found in the literature, the majority do not suggests a user-centered design, once virtual challenges were not tailored to the actual user’s motor needs, which can limit the clinical outcomes. Moreover, the studies analysed generally use virtual challenges aimed at entertainment with more playful activities, with a lack of ADLs. Therefore, there is space in future research to compare the efficacy of post-stroke balance rehabilitation of VR-based tools specifically designed for this purpose with existing VR-based tools. All the studies reported clinical outcomes, being the BBS and TUG the most used ones. They allow assessing static and dynamic balance that is usually affected after a stroke [47], [51]. 2.4 Conclusions This review focuses on the specifications of VR-based tools for post-stroke balance rehabilitation. The optimal VR tool solution should consider the requirements in terms on technology and balance rehabilitation. Thus, for the VR tool to have a user-centered design, it should consider the following technological requirements: 1) ensure wearable VR technology and biosensors in order to allow the performance of static and dynamic motor tasks; 2) identify the virtual challenges according to the user’s needs in daily activities, as well as the motor tasks that should be included in them; 3) ensure multimodal feedback with visual, auditory, and vibrotactile stimuli, which does not exist in the analysed literature, and 19 is needed to have a more realistic experience. Attending to the requirements of balance rehabilitation, the VR tool should: 1) select the motor tasks based on the most performed tasks in the literature, as well as the tasks included in the most performed clinical scales in the literature (BBS and TUG), these being: wight-shifting, stepping, look over the shoulder, reaching, knee flexion and extension, trunk and pelvic/hip movement, standing upright on one leg, sit-to-stand and stand-to-sit, and walking; 2) assess balance using key metrics, namely the position and velocity of the COM. 26 Table 3.3 - Virtual challenges and respective motor tasks Virtual challenges Motor tasks Fruit Catcher Cooking Take a Shower Watch TV Walking x Stepping x Sit-to-stand and stand-to-sit x Knee flexion and extension x x Weight shifting x x Trunk and pelvic /hip movement x Look over the shoulders x Stand upright on one leg x Reaching x The real-time development 3D platform Unity, created by Unity Technologies (Copenhagen, Denmark), with Editor version 2019.4.29f1 and Unity Hub 2.4.2 version, was used to create and develop the four 3D virtual challenges and all the surrounding VRE. It consists of a house with detailed indoor and outdoor spaces designed to be the most realistic and immersive as possible so that the users train according to their real needs and focus on the enthusiastic virtual challenge, respectively. To fulfil these purposes, in addition to the objects downloaded to fill the space and make it as welcome and real as possible, natural ambient sound and ambient sunlight were also created and designed to change the volume audio (low and high) and intensity of light (higher and lower) depending on whether the user is indoors or outdoors, respectively. Furthermore, the interior of the house was illuminated with the light from the sun coming through the windows without the need for light from lamps. In the remaining 27 subchapter, the four virtual challenges will be described (their main virtual goals and how to achieve them by practicing the related motor tasks) as well as their respective surrounding VRE. Fruit Catcher For the Fruit Catcher virtual challenge, it was created a backyard at the back of the house. In addition to the surrounding VRE (Figure 3.4), the backyard has a tree, a ladder, and apples as main virtual objects to carry out the virtual game (Figure 3.5). Figure 3.5 - Apple tree with the ladder. The Fruit Catcher game aims that the user: 1) climbs six steps of the ladder perched on the tree through stepping with each foot alternately; and 2) reach and catch nine apples, three forwards and three to each side (left and right), by performing weight shifting (with their feet together and without moving them) and looking forwards and over the left and right shoulders, respectively. The nine apples were placed at predefined distances (5 cm, 12.5 cm, and 25 cm after the hand position with arms Figure 3.4 - House backyard design. 28 outstretched, of a person with 1.70 m height) corresponding to the second, third, and fourth levels of the eighth BBS’ task (entitled “Reaching forward with outstretched arm while standing”), respectively. Cooking For the Cooking virtual challenge, a kitchen division in the house was created, with furniture, dishes, fridge, stove, and other electrical equipment (Figure 3.6). The four shelves full of food products (Figure 3.7) are the main virtual objects to carry out this virtual game. Figure 3.6 - Kitchen division design. Figure 3.7 - Kitchen shelves. The Cooking game aims that the user, with their feet together and without moving them, reaches the ingredients (selected to make a cake) at specific points of the shelves: 1) three levels of height adjusted according to user’s height (2 ingredients/level) through knee and hip flexion and extension, respectively; 2) and, on the left and right sides (3 ingredients/side) by performing weight shifting. The user should grab one of each of the ingredients tagged with labels (chocolate, eggs, flour, sugar, oil, and milk) and put them in the basket beside her/him, maintaining her/his balance. This virtual challenge was based on the activity of cooking, which is very often performed in a daily basis. 29 Take a Shower For the Take a Shower virtual challenge, a bathroom division was created in the house. In addition to the surrounding environment (Figure 3.8), a bathtub with a height of 40 cm (standard height of an undermount real bathtub [68]) (Figure 3.9) was implemented as being the main virtual object to carry out this game. Figure 3.8 - Bathroom division design. Figure 3.9 - Bathtub used in the "Take a Sower" game. The Take a Shower game aims that the user: 1) enters the bathtub with one foot at a time through knee flexion and extension without listening a sound of a kicking that is generated every time she/he touches the bathtub’s rim; 2) once inside the bathtub, stands upright on one leg, for listening to a sound of running water (auditory feedback) in case of success in lifting the leg enough, mimicking that she/he is washing her/himself. The user must lift the leg at least 25 cm and stand upright on one leg for at least 10 s (time to reach the maximum score on fourteenth BBS’s task: standing on one leg) to achieve the maximum score. This virtual challenge was based on the ADL of taking a bath, as it is an activity that is often performed and with a higher risk of falling. Thus, the motor task of getting into the bath and, inside 30 the bathtub, having some balance (e.g., when lifting one leg to wash it) should be trained so as not to fall in it. Watch Tv For the Watch Tv virtual challenge, a living room division in the house was created with a surrounding environment decorated as real as possible (Figure 3.10) and with special emphasis on the television and the sofa once they are the main virtual objects to carry out this game. The sofa was designed with a height of 47 cm according to the standards of real chairs [69]. Figure 3.10 - Living room division design. The Watch Tv game was inspired by the TUG clinical test. In this game, the user, initially raised, must: 1) sit in the virtual sofa (supported by a real matched chair); 2) stand; 3) walk 3 meters forward towards the television until it turns on; 4) rotate 180 degrees; 5) walk again 3 meters towards the sofa, ending the game. A video of nature with sound incorporated (visual and auditory feedback) plays on the television when the user reaches the 3 meters forward the sofa. The sit-to-stand and stand-to-sit tasks are performed in a different order when compared to the real TUG to ensure the user’s safety by matching the position of the real chair with the virtual sofa. 3.5 Control strategies This thesis developed control strategies to create a dynamic virtual environment that allows the users to interact with it and receive feedback, potentiating users’ motivation and participation during training and, thus, accelerating the recovery process. An avatar mimicking in real-time the user’s body movement was inserted in all the serious games, so that the virtual challenge seems realistic, and the user feels a high sense of presence in the VRE. Furthermore, the provided feedback (visual, auditory, or vibrotactile) allows the users to follow in real-time their postural control and to understand if the virtual challenge was successfully accomplished or not. 31 Feedback on balance indicators is relevant for postural control. Balance is maintained by regulating the interactions between the COM and base of support (BOS) [70]. In a brief description, in the context of defined motor tasks, the BOS is the area that includes every point of contact (one or both foots) that the person makes with the supporting surface (floor) and, in the case of having more than one point contact (both foots), the area between them too. The COM is the unique position where the sum of the weighted position vectors of all the parts of a system is equal to zero. Stable gait and balance are achieved if the COM representation on the supporting surface is within the BOS, being the limits of the BOS considered the limits of a stable balance [70]. For all the virtual challenges, in addition to a firstperson view, the user has access to a second real-time view regarding her/his COM projected on her/his feet. 3.5.1 First and second person views MVN Live Animation is a module, imported from the Unity Asset Store, in Unity software, that allows Unity developers to receive, process, and view Xsens live motions on an avatar (from “MvnPuppet” Prefab, Figure 3.11) [71]. The avatar is formed by 23 linked segments that move according to real-time streamed data (through the configuration of the Network Streamer mentioned in the subchapter 3.2) from MVN Analyze, mimicking the user’s movement. Figure 3.11 - Mvn avatar facing the Z-axis in Unity environment [71]. The first-person perspective was provided to allow high immersion, giving the user a sense of presence in the VRE, once she/he can interact with the virtual objects having a similar perspective to the one in the real world. This perspective was implemented by disabling the position tracking from the headset and controllers ("Track Pose Driver”) and inserting and adjusting them on avatar’s head ("Head_End”) and hands (“LeftCarpus 1” and “RightCarpus 1”), respectively. In this manner, the play 32 area moves according to the user’s motion in position and rotation (measured by Xsens sensors), allowing to control the inputs and outputs from headset and controllers. A second-person perspective view was created so that users can perform the virtual challenges and easily look at her/his feet with the COM projected. Therefore, this view shows the image provided by a camera pointed to the avatar’s feet (and that moves according to it, displaying the feet always pointing forward), providing visual feedback in real-time of simultaneously the user’s COM and BOS (defined by the area delimited by the feet on ground [70]). Moreover, the second-person view was associated with a plane that moves according to the avatar’s head so that, wherever the user looks, she/he can follow her/his BOS and projected COM, being able to adjust her/his posture during the practice of any virtual challenge towards maintaining balance and preventing her/him from falling. A representation of the COM (one red ball with a diameter of 1 cm) was implemented and placed at feet level to easily visualize if the COM is within the BOS. The position of the virtual red ball was controlled according to the streamed COM data (from MVN Analyze) post-transformed from the MVN reference axis to the Unity reference axis. A) B) Figure 3.12 – Watch Tv virtual challenge: A) In first person view; B) In firs person view, with the second real-time view (highlighted by the orange square). 3.5.2 Grab motion Both the apples on the tree and the food placed on the shelves (Fruit Catcher and Cooking games, respectively) are reached and caught using the HTC controllers. This control strategy allows the user to grab objects as long as they are not static objects and have physical properties. This function was 33 associated to both left and right controllers’ Trigger button by configuring “Open binding UI” to add “Grab” action on “SteamVR Input” (Figure 3.13 [72]). Figure 3.13 - HTC controllers' buttons and respective caption [72]. 3.5.3 Vibrotactile feedback Vibration was enabled in the HTC controllers to increase the immersion of the user during two situations: 1) when grabbing; 2) and colliding with virtual objects. This first situation was associated to both left and right controllers by configuring “Open binding UI” to add “Haptic” action associated with Trigger button. The haptic amplitude, duration, and frequency were set to 100 g, 0.4 s, and 22 Hz, respectively, when grabbing virtual objects). The second situation occurs when the controllers collide with virtual objects by enabling the controllers’ and virtual objects’ “Box Colliders” and “OnTriggerEnter” components. In this sense, every time a collider from a virtual object enters the controllers’ trigger, vibration is enabled (only in the controller that touches the virtual objects) with haptic amplitude, duration, and frequency of 88 g, 0.4 s, and 22 Hz, respectively, mimicking the sensation of hitting real objects with the hands. The amplitude, duration, and frequency values for two situations were selected in empirical tests with healthy subjects. 3.5.4 Visual and auditory feedback In the Watch Tv game, a video (inserted in a Video Player component) plays when the user approaches the TV and reaches the three-meter walk and stops 90 s after being activated, giving users’ interaction with the VRE, as if the user had turned the tv on. The Tv screen render texture was chosen to be the Target Texture where the video plays. The video plays showing images and sounds of nature when the avatar’s collider enters de television’s collider (trigger). 34 In the first part of the “Take a Shower” game, the user must enter the bathtub without touching the bathtub’s rim. However, a sound imitating a kick against a wall is enabled every time she/he touches it, as if she/he had really kicked the bathtub’s rim. To implement this control strategy, colliders were created both on the stone of the bathtub and on the user’s right and left feet. Whenever the collider of one of any user’s feet enter the collider of the bathtub’s rim (trigger), the sound plays, giving auditory feedback to user that she/he kicked the virtual bathtub’s rim. In the second part of the “Take a Shower” game, a sound of running water plays if the user stands upright on one leg inside the bathtub, as if she/he was washing the leg. The sound plays while the height of user’s foot (any foot) is above 25 cm from floor and stops 10 s after it starts (according to the fourteenth BBS’ task: “Standing on one leg”) or if the user puts the height of her/his foot below 25 cm from the floor. The height of the feet is assessed through “Position + Orientation (Quaternion)” streamed data from MVN Analyze. 3.5.5 Stepping up the ladder In the first part of the “Fruit Catcher” game, the user must climb the ladder perched in the tree by stepping one leg at a time. To develop this control strategy, the vertical and horizontal position of the avatar moves up and forward 25 cm and 4 cm (coinciding with the steps of the ladder), respectively, every time the height of any foot is above 25 cm from the floor and then backs to the floor’s height, mimicking that the user is climbing real stairs. 3.5.6 Shelves height To make the “Cooking” virtual challenge more user-centered and thus make the difficulty level as equivalent as possible between subjects according to their anthropometric measurements, the height of the shelves was automatically customised according to the user height. This personalisation was made for three shelves at different heights: the highest at eye level, the middle one at waist level, and the lowest at ankle level. Table 3.4 shows the height of the counter type furniture according to the users’ height. This information was used to personalise the height of the shelve at the waist level. These measures were found on a research of standard measures of kitchen shelves according to ergonomic rules [73]. 35 Table 3.4 - Counter type furniture’s height (cm) according to the user’s height (cm) User’s height (cm) Counter type furniture’s height (cm) 150 - 160 70 - 85 160 - 170 85 - 95 170 - 180 90 - 105 180 - 190 95 - 110 According to ergonomic rules, the shelf placed at eye level must be between 40 cm to 70 cm (for user’s height ranging from 150 cm to 190 cm, respectively) above the counter type furniture (shelf at waist level) [73]. Additionally, the shelf at ankle level must be located between 10 cm and 20 cm (for an user’s height ranging from 150 cm to 190 cm, respectively) from the floor level [73]. A fourth shelf, for decorative purposes only, was added as many centimetres above the eye-level shelf as the centimetres that the eye-level shelf is above the middle shelf. 3.6 Conclusions The fully immersive developed VR-based tool for balance training was designed in Unity 3D software, integrating the commercial setup device of VR HTC Vive Pro Full Kit to immerse the user in the VRE and the IMU-based motion capture system Xsens MVN Awinda, in a full-body configuration, to allow the user to interact with the VRE according to her/his motion. This is a wearable setup that enables balance training both in clinical and home settings. The VR-based tool includes four virtual challenges (corresponding to four ADLs) namely Fruit Catcher, Cooking, Take a Shower and Watch Tv, addressing a total of nine motor tasks (based on literature review, and BBS and TUG clinical tests) such as walking, stepping, sit-to-stand and stand-to-sit, knee flexion and extension, weight shifting, trunk and pelvic/hip movement, look over the shoulders, stand upright on one leg, and reach with hands as far as possible. To create the static VRE, a garden with an apple tree with a perched on it, a kitchen with shelves with food, a bathroom with a bathtub, and a living room with a sofa and a Tv were created for the Fruit Catcher, Cooking, Take a Shower, and Watch TV virtual challenges, respectively. To create a dynamic virtual environment and allow the user to interact with it and receive feedback from it, six control strategies were created, namely first and second person views, grab, vibrotactile feedback, visual and auditory feedback, stepping up the ladder, and shelves height. The tool provides visual, auditory and vibrotactile stimuli through HMD, the built-in headphones, and HTC controllers, respectively. Future work envisages the creation of an interactive menu. 42 Figure 4.6 - Number of subjects by gender (female and male), in percentage, by activity, considering the five activities, out of fifteen, that people do the most. Figure 4.7 - Number of subjects by gender (female and male), in percentage, by activity, considering the five activities, out of fifteen, that people like to do the most. Figure 4.6 shows that the five activities most performed by females are taking a shower (11.67 %), watching television, and climbing the stairs in the house/building with the same percentage (9.46 %), followed by cleaning the floor and driving the car, with the same percentage too (5.99 %). On the other Dance Take a shower Cook Walking a pet Clean house windows Clean the floor Ride public transport Riding a bicycle Driving a car Playing cards Watch television Climb a tree and pick fruit Change the burned-out lamps Tightening the laces Climb the stairs in the house/building 0.32 2.52 6.31 11.67 10.09 0.95 0.00 2.52 2.52 2.21 5.99 0.00 0.63 0.63 1.58 1.89 4.73 5.99 0.00.032 5.36 9.46 0.00 0.00 3.47 4.73 9.46 0.00 2.00 6.62 4.00 6.00 8.00 10.00 12.00 14.00 Number of people (%) Female Male Dance Take a shower Cook Walking a pet Clean house windows Clean the floor Ride public transport Riding a bicycle Driving a car Playing cards Watch television Climb a tree and pick fruit Change the burned-out lamps Tightening the laces Climb the stairs in the house/building 1.26 5.66 9.12 9.75 3.14 3.14 7.86 6.29 0.00 1.57 0.63 1.57 0.31 1.57 3.774.40 4.405.03 3.14 5.66 5.97 8.49 0.31 0.00 2.52 0.63 0.00 0.94 1.57 1.26 2.00 4.00 6.00 8.00 10.00 12.00 Number of people (%) Female Male Activities Activities 43 hand, the male gender reports performing more tasks such as climbing the stairs in the house/building (6.62 %), taking a shower (6.31 %), watching television (5.36 %), followed by driving a car and tightening the laces with the same percentage (4.73 %). Regarding the five activities most appreciated by female gender, presented in Figure 4.7, these are taking a shower (9.75 %), dancing (9.12 %), watching television (8.49 %), cooking (7.86 %), and walking a pet (6.29 %). The male gender, on the other hand, prefers activities such as watching television (5.97 %), taking a shower (5.66 %), driving a car (5.03 %), riding a bicycle (3.77 %), followed by cooking, walking a pet, and playing cards with the same percentage (3.14 %). 4.4.4 Degree of daily physical activity From the participants, 67.19 % consider themselves to be an active person. The Figures 4.8 and 4.9 present the results from the subjects by degree of daily physical activity about the five ADLs they most perform and most enjoy doing, respectively. Figure 4.8 - Number of subjects by degree of daily physical activity (active and not active), in percentage, by activity, considering the five activities, out of fifteen, that people do the most. Dance Take a shower Cook Walking a pet Clean house windows Clean the floor Ride public transport Riding a bicycle Driving a car Playing cards Watch television Climb a tree and pick fruit Change the burned-out lamps Tightening the laces Climb the stairs in the house/building 1.26 1.58 5.99 11.99 8.52 1.89 1 4.10 .58 0.32 1.89 5.36 0.91.58 2.21 5 0.00.63 0.32 0.00 2.52 8.20 5.68 9.15 0.00 0.00 2.52 2.00 4.00 5.68 5.36 6.00 10.73 0.00 8.00 10.00 12.00 14.00 Number of people (%) Active Not active Activities 44 Dance Take a shower Cook Walking a pet Clean house windows Clean the floor Ride public transport Riding a bicycle Driving a car Playing cards Watch television Climb a tree and pick fruit Change the burned-out lamps Tightening the laces Climb the stairs in the house/building 3.77 6.60 4.72 10.69 3.77 7.23 3.4 6 5.97 0.94 0.63 1.89 0.31 26 1. 0.63 2.52 5.66 2.83 6.60 3.4 6 5.3 5 4.40 10.06 2.20 0.63 0.00 0.94 0.63 2.20 0.63 0.00 2.00 4.00 6.00 8.00 10.00 12.00 Number of people (%) Active Not active Figure 4.9 - Number of subjects by degree of daily physical activity (active and not active), in percentage, by activity, considering the five activities, out of fifteen, that people like to do the most. Considering the Figure 4.8, it is possible to verify that the five activities most performed by physically active persons are taking a shower (11.99 %), climbing the stairs of the house/building (10.73 %), watching television (9.15 %), cooking (8.52 %), and driving a car (8.20 %). Regarding the not physically active people, the answers are taking a shower (5.99 %), watching television (5.68 %), climbing the stairs of the house/building (5.36 %), and cooking (4.10 %), followed by driving a car and tightening the laces with the same percentage (2.52 %). On the other hand, when choosing the five ADLs they most like to do (Figure 4.9), physically active people chose taking a shower (10.69 %), watching television (10.06 %), and cooking (7.23 %), followed by dancing and driving a car on the same level (6.60 %). The not physically active people selected taking a shower (4.72 %), watching television (4.40 %), dancing and cooking with the same value (3.77 %), followed by walking a pet and playing cards with the same percentage (3.46 %). 4.4.5 Health condition People were asked whether they had any illness that prevented them from being more active or not. In this way, and according to the answers, 90.62 % of population was healthy and the remaining 9.38 % was referred as injured subjects. Figures 4.10 and 4.11 represent the most performed and enjoyed activities in people’s daily life, respectively, by health condition. Activities 45 Figure 4.10 - Number of subjects with and without disease, in percentage, by activity, considering the five activities, out of fifteen, that people do the most. Figure 4.11 - Number of subjects with and without disease, in percentage, by activity, considering the five activities, out of fifteen, that people like to do the most. Analysing the Figure 4.10, the five activities most performed by healthy people (without disease) are taking a shower (16.72 %), climbing the stairs of the house/building (15.14 %), watching television (13.25 %), cooking (11.36 %), and driving a car (10.09 %). On the other hand, people with a disease tend to do more activities such as cleaning the floor and watching television on the same level (1.58 %), Dance Take a shower Cook Walking a pet Clean house windows Clean the floor Ride public transport Riding a bicycle Driving a car Playing cards Watch television Climb a tree and pick fruit Change the burned-out lamps Tightening the laces Climb the stairs in the house/building 0.00 2.84 1.26 1.26 0.32 0.631.58 16.72 11.36 3.15 5.99 0.00 1.58 2.52 0.000.63 10.09 0.32 0.00 0.63 1.58 13.25 0.00 0.00 0.63 0.95 2.00 7.57 15.14 0.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 Number of people (%) Without disease With disease Dance Take a shower Cook Walking a pet Clean house windows Clean the floor Ride public transport Riding a bicycle Driving a car Playing cards Watch television Climb a tree and pick fruit Change the burned-out lamps Tightening the laces Climb the stairs in the house/building 0.63 1.26 0.94 0.31 0.94 9.75 14.15 10.06 9.12 0.63 0.31 1.89 0.31 1.57 0.00 0.94 0.63 1.26 8.18 8.49 8.18 13.21 0.63 0.00 0.94 2.20 0.00 0.63 0.31 2.00 2.52 4.00 6.00 8.00 10.00 12.00 14.00 16.00 Number of people (%) Without disease With disease Activities Activities 46 followed by taking shower and cooking (1.26 %), and finally climbing the stair of the house/building (0.95 %). Figure 4.11 shows that the five activities most appreciated by the healthy population (without disease) are taking a shower (14.15 %), watching television (13.21 %), cooking (10.06 %), dancing (9.75 %), and walking a pet (9.12 %). People with illness, on the other hand, prefer activities such as taking a shower and watching television with the same percentage (1.26 %), cooking and driving a car at the same level too (0.94 %), followed by dancing, cleaning house windows, playing cards, and climbing a tree to pick fruit with the same percentage (0.63 %). 4.4.6 Degree of home urbanization This subchapter aims to study if the urbanization of the place where the person lives influences the activities that they do the most and the ones they like to do the most. From the participants, 81.25 % live in the city. Figures 4.12 and 4.13 present the most performed and appreciated activities, respectively, by the subjects according to the degree of urbanization where they live (city or countryside). Dance Take a shower Cook Walking a pet Clean house windows Clean the floor Ride public transport Riding a bicycle Driving a car Playing cards Watch television Climb a tree and pick fruit Change the burned-out lamps Tightening the laces Climb the stairs in the house/building 0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 Number of people (%) City Countryside Figure 4.12 - Number of subjects living in the city and in countryside, in percentage, by activity, considering the five activities, out of fifteen, that people do the most. Activities 2.52 0.32 2.52 15 .46 2.52 10.09 3.1 5 0.32 1.89 0.32 1.58 5.99 2.52 0.00 0.32 0.32 2.21 8.52 0.32 0.00 3.79 11. 04 0.00 0.00 1.58 6.62 3. 47 12.6 2 47 Figure 4.13 - Number of subjects living in the city and in countryside, in percentage, by activity, considering the five activities, out of fifteen, that people like to do the most. Considering the Figure 4.12, the five most frequently performed activities by people living in the city are taking a shower (15.46 %), climbing the stairs in the house/building (12.62 %), watching television (11.04 %), cooking (10.09 %), and driving a car (8.52 %). People living in the countryside most often perform activities such as watching television (3.79 %), climbing the stairs in the house/building (3.47 %), taking a shower and cooking (2.52 %), followed by cleaning the floor and tightening the laces with the same percentage value (1.58 %). Regarding the Figure 4.13, the five activities most enjoyed by people living in the city are taking a shower (12.89 %), watching television (10.69 %), dancing (9.12 %), cooking (8.81 %), and walking a pet (8.18 %). On the other hand, people living in the countryside, tend to enjoy more activities such as watching television (3.77 %), taking a shower and driving a car with the same percentage (2.52 %), cooking and playing cards also with the same percentage (2.20 %). 4.4.7 Type of transport daily used The transport daily used by the population may influence the activities that people most perform and most like to perform in their daily lives. The number of individuals, in percentage, who commute by motorised or non-motorised transport is 63.75 % and 36.25 %, respectively. The answers of the subjects about the five activities they most perform and enjoy in their daily lives by type of transport daily used is presented in Figures 4.14 and 4.15, respectively. Dance Take a shower Cook Walking a pet Clean house windows Clean the floor Ride public transport Riding a bicycle Driving a car Playing cards Watch television Climb a tree and pick fruit Change the burned-out lamps Tightening the laces Climb the stairs in the house/building 1.26 9.12 2.52 2.20 12.89 1.26 0.31 1.26 8.81 8.18 0.00 0.00 2.20 1.89 1.57 2.52 2.20 6.60 6.92 6.60 3.77 10.69 0.31 0.00 0.94 2.52 0.63 0.31 2.00 2.52 0.00 4.00 6.00 8.00 10.00 12.00 14.00 Number of people (%) City Countryside Activities 48 Dance Take a shower Cook Walking a pet Clean house windows Clean the floor Ride public transport Riding a bicycle Driving a car Playing cards 1.51 1.76 6.55 11.34 4.03 8.06 0.76 2.27 0.761.01 3.02 4.03 1.011.76 0.25 0.25 0.25 0.50 2.27 8.56 9.32 Tightening the laces Climb the stairs in the house/building 5.04 10.83 0.00 2.00 4.00 5.79 6.00 8.00 10.00 12.00 Number of people (%) Motorised transport Non-motorised transport Watch television 5.29 Climb a tree and pick fruit 0.00 Change the burned-out lamps 0.00 3.78 Figure 4.14 - Number of subjects who commute by motorised or non-motorised transport, in percentage, by activity, considering the five activities, out of fifteen, that people do the most. Figure 4.15 - Number of subjects who commute by motorised or non-motorised transport, in percentage, by activity, considering the five activities, out of fifteen, that people like to do the most. Analysing the Figure 4.14, it is possible to verify that people who move more often through motorized transport (by car or public transport) do more activities such as taking a shower (11.34 %), climbing the stairs in the house/building (10.83 %), watching television (9.32 %), driving a car (8.56 %), and cooking (8.06 %). People who move more often through non-motorised transport (such as on foot) Dance Take a shower Cook Walking a pet Clean house windows Clean the floor Ride public transport Riding a bicycle Driving a car Playing cards Watch television Climb a tree and pick fruit Change the burned-out lamps Tightening the laces Climb the stairs in the house/building 4.52 5.03 4.02 6.03 9.80 7.29 6.03 0.50 1.01 3.02 1.01 0.75 1.01 1.01 2.76 2.01 3.02 6.28 6.78 6.03 5.28 9.55 0.00 1.01 1.51 1.01 0.75 1.51 1.51 0.00 2.00 4.00 6.00 8.00 10.00 12.00 Number of people (%) Motorised transport Non-motorised transport Activities Activities 49 tend to perform more frequently activities such as taking a shower (6.55 %), climbing the stairs of the house/building (5.79 %), watching television (5.29 %), cooking (4.03 %), and tightening the laces (3.78 %). On the other hand, regarding the most appreciated activities (Figure 4.15), those who commute most often by motorised transport choose taking a shower (9.80 %), watching television (9.55 %), cooking (7.29 %), driving a car (6.78 %), and riding a bicycle (6.28 %). People who move more often by nonmotorised transport tend to prefer activities such as watching television (5.28 %), taking a shower (5.03 %), dancing (4.52 %), and cooking (4.02 %), followed by walking a pet, and playing cards with the same percentage (3.02 %). 4.4.8 Conventional physical therapy Figure 4.16 shows the number of participants who have never or ever had undertaken conventional physical therapy and, if so, whether they felt motivated or not during the sessions. Figure 4.16 - Number of the individuals, in percentage, who do, did, or did not do physical therapy and whether they felt/feel or not motivated during sessions. According to Figure 4.16, it is possible to verify that 53.13 % of participants never performed conventional physical therapy. Of the remaining subjects who have already done or are doing conventional physical therapy (46.87 %), 31.24 % felt or feel motivated during sessions. For those who do or have done conventional physical therapy and did not feel motivated (15.63 % of the participants), they were asked to indicate the reasons for such a statement. The answers of these subjects are presented in Figure 4.17. Question: Do you do or have you done physical therapy? Did you feel motivated during the sessions? 15.63% 53.13% 31.24% No Yes, and I feel/felt Yes, and I do not feel/felt 50 Reasons of unmotivated people about physical therapy Tiring Boring Demanding Routine Lack of monitoring by a physiotherapist 0 10 20 30 40 50 60 70 80 90 100 Number of people (%) Figure 4.17 - Number of people, in percentage, among those who do not felt/feel motivated, by reasons (tiring, boring, demanding, other) for such discontentment about conventional physical therapy. 4.5 Discussion This questionnaire intends to study what are the five ADLs, among fifteen options given, that people do the most and the ones they most enjoy doing to create a VR-based tool for balance training, as usercentered as possible. On the one hand, it is important to focus on the activities that the user does most in her/his daily life so that the training has a direct impact on her/his life. On the other hand, it is also relevant to focus the training on activities that the user enjoys doing so that rehabilitation motivates the users and does not become tedious and boring. In this manner, the user tends to actively participate in the training, accelerating the recovery process. Thus, it is important to find a middle ground between what the user does most and what the user likes to do the most to take advantage of both types of activities. This questionnaire is intended to check whether the tool created in this project is in line with a user-centered design and what can be done to improve it, guiding future research. According to the global results from the questionnaire (Figures 4.1 and 4.2), there are four simultaneously most performed and liked activities: taking a shower, watching television, cooking, and driving a car. Climbing the stairs of the house/building is also one of the most performed activities; however, it is not in the subjects’ preferences. This happens probably because it is an activity that people mostly perform for the need to access other places and does not contain any kind of pleasure. On the other hand, dancing and walking a pet are two of the most appreciated activities, although they are not Reasons 51 the most performed since most of the studied population has an active labour life. Overall, it is possible to verify that there is an agreement in four out of five activities, which indicates that the most performed activities are also among the most appreciated, and vice versa. Comparing the four activities that were simultaneously most performed and appreciated with the four virtual challenges created in this dissertation, it is possible to verify that there are three activities in common (taking a shower, watching television, and cooking), indicating that the developed VR tool follows the needs and the preferences of the participating population. Climbing stairs is the fourth motor task most performed by participants, which is included in the Fruit Cather virtual challenge. Analysing the activities according to the subjects’ age (Figures 4.4 and 4.5), despite the wide range of ages among the subjects, taking shower, climbing the stairs of the house/building, watching television, and cooking are four of the most frequently performed ADLs in common between the three age groups analysed. Driving a car is also one of the most performed activities in subjects belonging to the first two age groups (16-40 years and 40-66 years), and cleaning the floor is one of the most performed activities among the last two age groups (40-66 years and > 66 years). Thus, it is possible to determine that age has low influence on the tasks that people do more on a daily basis. However, since aging is more prone to diseases that limit balance, the age presents a decline in the functional status of patients being a common cause for limitations in ADLs [75]. Thus, it is normal that these people perform different ADLs than younger and healthier people. These results may be negatively influenced by the fact that the study population does not present balanced groups. Regarding the ADLs most appreciated by the subjects, the scenario changes. Watching television and dancing are the activities that satisfy the largest number of individuals regardless of age. Future research following a user-centered design should have in consideration that the age groups feel motivated with different ADLs. Considering the gender factor, four of the five most performed activities are common to both genders (taking shower, climbing the stairs of the house/building, watching television, and driving a car, Figure 4.6). Regarding the most appreciated ADLs, among the choices made by both groups, there are also four activities in common (taking a shower, watching television, cooking, and walking a pet, Figure 4.7). These results suggest that gender has a low influence on the most performed and appreciated ADLs. Diving the population into active and not active persons, both groups perform the most the same type of activities (taking a shower, climbing the stairs of the house/building, watching television, cooking, and driving a car, Figure 4.8). Regarding the five activities most appreciated by both groups, there are four in common (taking a shower, watching television, dancing, and cooking, Figure 4.9). Thus, the results 58 with a maximum score of 6.44. For the IMI, intended to assess the participants’ subjective experience related to the VR experience lived by them during the VR training [77], the subscales to be analysed were chosen according to the ones that best fit with the activity performed: Interest/Enjoyment, Perceived Competence, Effort/Importance, Pressure/Tension, and Value/Usefulness. The questionnaire had a total of 30 questions, of which 27 were rated on a scale of 1 to 7, with the remaining 3 questions being openended. Thus, each of the subscales presents the best score for 7.00, 7.00, 2.20, 1.00, and 7.00, respectively. For the third and fourth subscales, the lower and closer to the minimum value (2.20 and 1.00, respectively) the better, contrary to the remaining subscales that the higher the score the better. 59 Table 5.2 – Virtual challenges and their respective outcome measures Virtual Challenge Motor Tasks Sensor-based Outcomes Virtual Challenges Scores Fruit Catcher 1. Stepping 2. Weigh shifting 3. Looking over the shoulders 4. Reaching Primary During climbing stairs: • COM’s position on the AP and ML directions • COM’s velocity on the AP and ML directions During catching fruits: • COM’s position on the AP and ML directions • COM’s velocity on the AP and ML directions - Number of steps climbed - Number of apples caught Secondary • Knee joint angle on the sagittal plane • Hand position to the left, right, and forward directions Take a Shower 1. Knee flexion/extension 2. Stand upright on one leg Primary While entering in the bathtub: • COM’s position on the AP and ML directions • COM’s velocity on the AP and ML directions Inside the bathtub: • COM’s position on the AP and ML directions • COM’s velocity on the AP and ML directions - Number of times the user kicked the bathtub rim - Sound playing duration inside bathtub Secondary • Knee joint angle on the sagittal plane Cooking 1. Knee flexion/extension 2. Weigh shifting 3. Trunk and pelvic/hip movement Primary • COM’s position on the AP and ML directions • COM’s velocity on the AP and ML directions - Number of ingredients caught in kitchen’s shelves Watch Tv 1. Stand-to-sit 2. Sit-to-stand 3. Walking Primary Sit-to-stand: • COM’s position on the AP and ML directions • COM’s velocity on the AP and ML directions Stand-to-sit: • COM’s position on the AP and ML directions • COM’s velocity on the AP and ML directions During walking: • COM’s position on the AP and ML directions • COM’s velocity on the AP and ML directions - Television state achieved Secondary • Knee joint angle on the sagittal plane • Foot contact 60 Table 5.3 – Clinical scales and their respective outcome measures Clinical Scale Motor Tasks Sensor-based Outcomes Clinical Scales Scores BBS 1. Standing unsupported 2. Sitting with back unsupported 3. Transfers 4. Standing unsupported with eyes closed 5. Standing unsupported with feet together 6. Pick up an object from the floor 7. Turn to look behind the shoulders 8. Turn 360 degrees 9. Standing unsupported one foot in front Primary • COM’s position on the AP and ML directions • COM’s velocity on the AP and ML directions - Score for each BBS’ task 10. Sitting to standing 11. Standing to sitting 12. Place alternate foot on step while standing unsupported 13. Standing on one leg Primary • COM’s position on the AP and ML directions • COM’s velocity on the AP and ML directions Secondary • Knee joint angle on the sagittal plane 14. Reaching forward with outstretched arms while standing Primary • COM’s position on the AP and ML directions • COM’s velocity on the AP and ML directions Secondary • Hand position to the left, right, and forward directions TUG 1. Sit-to-stand 2. Walking 3. Stand-to-sit Primary • COM’s position on the AP and ML directions • COM’s velocity on the AP and ML directions - Time to perform the TUG Secondary • Knee joint angle on the sagittal plane • Foot contact 61 5.4 Data processing The data processing was carried out using Matlab software (MATLAB R2020a, © 1994-2021 The MathWorks, Inc.). Firstly, data from the virtual challenges of Fruit Catcher, Take a Shower, and Watch Tv, and from the TUG test were segmented between the related motor tasks. For the Fruit Catcher and Take a Shower virtual challenges, there are two distinct activities to be analysed: climbing the stairs and catching fruits; and entering in the bathtub and standing on one leg while inside the bathtub, respectively. For the Watch Tv virtual challenge and the TUG clinical test, there are three different activities: sit-tostand, walking, and stand-to-sit. The segmentation performed was done by visual inspection of the kinematic data or by foot contact. Data from the Cooking virtual challenge and the BBS’ tasks were not segmented. Secondly, metrics were calculated from the sensor-based data. The COM displacement in the AP and ML directions was calculated by the difference between the highest and the lowest values of the COM’s position in the x and y axes, respectively. The maximum and minimum COM velocity in the AP and ML directions correspond to the highest and lowest values of the COM velocity in the x and y axes, respectively. The ROM of the left and right knee joint angle in the sagittal plane was determined by the difference between the highest and the lowest angle in the z-axis of the left and right knee joint, respectively. The maximum leftand right-hand position corresponds to the maximum leftand right-hand position on the y-axis, respectively. On the other hand, the maximum forward hand position corresponds to the highest value on the x-axis between the maximum positions of the left and right hands. Regarding spatiotemporal gait metrics (stride length and stride time, respectively), both were determined based on the right leg (participants’ dominant leg) foot contact. The stride length was determined by the difference of the right foot position, on the x-axis, between consecutive right foot contacts on the ground. Additionally, the stride time was calculated through the difference of the frames between consecutive right foot contacts on the ground and, then, divided by the sampling frequency (60 Hz). The metrics related to the walking motor task were normalized per gait cycle, making the subsequent calculation of mean and standard deviation (SD) for data presentation. 5.5 Statistical Analysis The statistical analysis was performed through IBM SPSS software version 26.0 (for Windows) (IMP Corp, Armonk, NY, USA). The data normality and homoscedasticity were assessed using KolmogorovSmirnov’s and Levene’s tests, respectively. Two-tailed t -tests were performed for the parametric metrics 62 considering the “time” factor. The Wilcoxon signed-rank test was applied to variables where at least one of the assumptions (normality or homoscedasticity) was not verified. All statistical tests were executed considering a confidence level of 95 % (α = 0.05) and paired samples. Only the primary outcomes were statistically analysed. The statistical tests were conducted to evaluate the following null hypotheses: 1) there are no statistically significant differences between trial 1 and trial 2 of VR intervention; 2) there are no statistically significant differences between before and after VR intervention. 5.6 Conclusions The six participants have successfully and safely completed the intervention, within 1h, which consisted of 1 VR training session with 1 VR familiarization trial and 2 data acquisition data trials in a row, per virtual challenge. The primary outcomes during VR intervention and the BBS and TUG clinical tests are the COM’s position and velocity in the AP and ML directions. Secondary outcomes included: 1) knee joint angle on the sagittal plane, hand position to the left, right and forward directions, and foot contact data, for both; 2) virtual challenges, during VR intervention; 3) clinical scales scores, for BBS and TUG. Both primary and secondary outcomes were well-collected during the intervention. Finally, all participants answered the IMI and IPQ user’s experience evaluation tests after the VR intervention (secondary outcome). Data processing functions were successfully implemented to determine the metrics of COM displacement, minimum and maximum COM velocity on the AP and ML directions, ROM of the knee joint angle on the sagittal plane, maximum hand position to the left, right, and forward directions, and spatiotemporal gait parameters (stride length and stride time, respectively). Further, the statistical analysis, including the two-tailed t -tests for parametric metrics and Wilcoxon signed-rank tests for nonparametric metrics, was carried out with success for the estimated metrics. 63 6. RESULTS AND DISCUSSION This chapter aims to present the results from the preliminary validation of the VR-based tool for balance training. The results are divided into three parts. First, all primary and secondary outcomes concerning the VR intervention are presented. The second part of this chapter presents the results of the sensor-based and clinical scales assessment performed before and after the VR intervention for the BBS and TUG related motor activities. Third, the answers of the user’s experience evaluation tests are shown. 6.1 During VR Intervention After processing the data, the tables with all the primary and secondary outcomes presented in Appendix I were obtained, as well as the figures presented in this subchapter, illustrating the COM displacement in the AP direction by the ML direction per each virtual challenge. 6.1.1 Fruit Catcher Figure 6.1 presents the COM displacement in the AP direction by the ML direction for all participants in each trial during the tasks of climbing stairs and catching fruits of the virtual challenge Fruit Catcher. Figure 6.1 - COM displacement (cm) in the AP direction by the ML direction, for all participants in each trial during the tasks of climbing stairs and catching fruits, of the virtual challenge "Fruit Catcher". 64 After the construction of the Tables 0.1 and 0.2 presented in Appendix I, the data was statistically evaluated, comparing the primary outcomes from T1 versus T2 for each virtual challenge. Table 6.1 presents the results of parametric (paired samples t-test) and non-parametric metrics (Wilcoxon signedrank test) for the virtual challenge Fruit Catcher. Table 6.1 - Mean and SD of T1 and T2, and p-value of the paired samples test and Wilcoxon test for COM displacement (Disp) and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for climbing stairs and catching fruits motor tasks of the virtual challenge “Fruit Catcher”, considering the time condition and a significance level of 5 %. The most representative results ( p -value < 0.05) appear in bold Paired Samples Test Test Statistics a Fruit Catcher Mean ± SD t -test Wilcoxon T1 T2 t Sig. (2tailed) Z Asymp. Sig. (2tailed) Climbing Stairs DispAP_CS 14.485 ± 2.531 12.437 ± 1.459 - - -1.153b .249 DispML_CS 17.868 ± 2.256 15.982 ± 2.507 2.069 .093 - - VelMinAP_CS - 5.593 ± 0.376 - 4.863 ± 0.321 -1.646 .161 - - VelMaxAP_CS 10.640 ± 1.392 10.063 ± 0.760 .483 .649 - - VelMinML_CS -20.152 ± 2.847 -20.047 ± 2.413 -0.100 .924 - - VelMaxML_CS 18.337 ± 1.649 16.882 ± 1.345 1.245 .268 - - Catching Fruits DispAP_CF 15.013 ± 1.701 16.015 ± 1.403 -.461 .664 - - DispML_CF 20.332 ± 2.948 22.113 ± 2.430 -1.379 .226 - - VelMinAP_CF -11.300 ± 0.652 -12.015 ± 1.148 .851 .434 - - VelMaxAP_CF 9.482 ± 0.687 9.358 ± 0.943 .173 .869 - - VelMinML_CF -14.210 ± 2.436 -13.285 ± 1.723 -.344 .745 - - VelMaxML_CF 14.323 ± 2.485 11.408 ± 1.574 1.530 .187 - - a. Wilcoxon Signed Ranks Test b. Based on positive ranks 65 Table 6.2 indicates the mean and SD of the secondary outcomes of all the participants, for each trial of the virtual challenge Fruit Catcher, namely the ROM of left and right knee joint angles on the sagittal plane (degrees) during climbing stairs, and the maximum hand position to the left, right, and forward directions (cm) during catching fruits. Table 6.2 – Mean and SD of the ROM of the left and right knee joint angles on the sagittal plane (degrees), and the maximum hand position to the left, right, and forward directions (cm) in each trial of the virtual challenge “Fruit Catcher” Secondary Outcomes Mean ± SD T1 T2 Climbing stairs ROM of the knee joint angles (degrees) Left knee 79.18 ± 7.49 79.18 ± 9.28 Right knee 77.38 ± 7.25 74.63 ± 6.81 Catching Fruits Maximum hand position (cm) Left 51.67 ± 6.21 51.38 ± 7.62 Right 69.87 ± 8.17 69.25 ± 7.44 Forward 86.18 ± 4.29 86.80 ± 6.47 Table 6.3 shows the virtual challenge scores for each trial of both the motor tasks, climbing stairs and catching fruits, of the virtual challenge Fruit Catcher. Table 6.3 – Number of steps climbed, and number of apples caught in each trial of climbing stairs and catching fruits, respectively, in virtual challenge “Fruit Catcher” Virtual challenge scores Participants Number of steps climbed Number of apples caught T1 T2 T1 T2 P1 6 6 6 6 P2 6 6 5 7 P3 6 6 7 6 P4 6 6 9 9 P5 6 6 4 4 P6 6 6 9 9 66 Although the statistical analysis performed to the virtual challenge Fruit Catcher did not indicate any significant difference between trials ( p -value > 0.05), the data of COM displacement in the ML direction presents a positive probably value ( t = 2.069) (Table 6.1). This indicates a general trend of its decreasing between trials. Since people continued to perform the intended activity exquisitely as they were able to climb all 6 stairs, a decrease in the COM value indicates that, in general, participants in the second trial were able to achieve the same goal by varying the COM less, which means increased stability of balance. As this motor activity involves a greater variation of COM in the ML direction than on the AP axis, an improvement in COM displacement in this direction (ML) is a very positive assessment. Although the lower variation of COM displacement in the AP direction, participants also decreased this metric (14.485 ± 2.531 cm on T1, and 12.437 ± 1.459 cm on T2) (Table 6.1). Considering this, it is possible to verify that despite the improvements at the balance level, there was no increase in COM velocity in any direction of any of the axes (mean and standard deviation in T1 is lower than in T2) (Table 6.1). Regarding the second motor task of catching fruits, there was only an increase in AP velocity in the negative direction of the axis (after picking a fruit, the user returned to the initial resting position) (Table 6.1). The no decrease in COM displacement in either direction (AP and ML) can be explained by the fact that participants in the second trial were trying to pick more fruit than in the first trial. This behaviour is shown by increasing the maximum hand position in forward direction (Table 6.2). Furthermore, with exception of P4 and P6, none of the other participants succeeded in picking the maximum number of apples, in both trials (Table 6.3), which triggers an interest in reaching further and picking more fruit. P2 was able to increase from 5 to 7 apples picked between T1 and T2 (Table 6.3). All of this can justify the non-decrease of COM displacement. 6.1.2 Take a Shower Figure 6.2 illustrates the participants’ COM displacement for AP direction by ML direction in each trial during the tasks of entering in the bathtub and standing on one leg inside the bathtub of the virtual challenge Take a Shower. 67 Figure 6.2 - COM displacement (cm) in the AP direction by the ML direction for all participants in each trial during the tasks of entering in the bathtub and standing on one leg inside the bathtub, of virtual challenge "Take a Shower". After the construction of the Tables 0.3 and 0.4 presented in Appendix I, the data was statistically evaluated, comparing the primary outcomes from T1 versus T2 for each virtual challenge. Table 6.4 presents the results of parametric (paired samples t -test) and non-parametric metric (Wilcoxon signedrank test) for the virtual challenge Take a Shower. 74 Table 6.9 - Mean and SD of T1 and T2,, and the p-value of the paired samples test and Wilcoxon test for COM displacement (Disp) and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for stand-to-sit, sit-to-stand, and walking motor tasks of the virtual challenge “Watch Tv”, considering the time condition and a significance level of 5 %. The most representative results ( p -value < 0.05) appear in bold Paired Samples Test Test Statistics a Watch TV Mean ± SD t -test Wilcoxon T1 T2 t Sig. (2tailed) Z Asymp. Sig. (2tailed) Stand-to-sit DispAP_Sit 35.080 ± 3.029 35.822 ± 2.398 -.367 .729 - - DispML_Sit 3.477 ± 0.551 4.532 ± 0.553 - - -1.153c .249 VelMinAP_Sit -24.963 ± 3.934 -23.975 ± 1.270 -.235 .824 - - VelMaxAP_Sit 1.793 ± 0.295 1.713 ± 0.456 .146 .890 - - VelMinML_Sit -5.540 ± 0.693 -6.422 ± 1.525 .576 .589 - - VelMaxML_Sit 4.333 ± 0.672 5.670 ± 1.419 - - -1.572c .116 Sit-to-stand DispAP_Stand 33.727 ± 3.080 33.950 ± 2.605 -.125 .906 - - DispML_Stand 3.373 ± 0.452 2.857 ± 0.414 1.281 .256 - - VelMinAP_Stand 0.897 ± 0.508 -0.740 ± 0.840 2.276 .072 - - VelMaxAP_Stand 32.327 ± 1.211 37.823 ± 3.182 - - -1.572c .116 VelMinML_Stand 3.860 ± 0.944 4.933 ± 1.012 -1.746 .141 - - VelMaxML_Stand -5.705 ± 0.878 -4.102 ± 0.670 -1.531 .186 - - Walking DispAP_Walk 73.708 ± 3.391 77.713 ± 5.760 - - -1.153c .249 DispML_Walk 9.113 ± 0.982 7.483 ± 0.553 2.587 .049 - - VelMinAP_Walk 42.647 ± 3.259 46.773 ± 3.836 -3.381 .020 - - VelMaxAP_Walk 64.854 ± 3.830 70.702 ± 4.675 -2.571 .050 - - VelMinML_Walk -20.338 ± 1.881 -20.137 ± 1.330 -.105 .921 - - VelMaxML_Walk 18.360 ± 3.261 17.202 ± 2.438 - - -.734b .463 a. Wilcoxon Signed Ranks Test b. Based on positive ranks c. Based on negative ranks 75 Table 6.10 shows the mean and SD of the secondary outcomes of all the participants, for each trial of the virtual challenge Watch Tv, namely, ROM of left and right knee joint angles on the sagittal plane (degrees) during stand-to-sit, sit-to-stand, and walking, and spatiotemporal gait parameters such as stride length (cm) and stride time (s) for walking motor task. Table 6.10 – Mean and SD of the ROM of the left and right knee joint angles on the sagittal plane (degrees) for stand-to-sit, sit-to-stand, and walking motor tasks, and the stride length (cm) and stride time (s) for walking motor task in each trial of the virtual challenge “Watch Tv” Secondary Outcomes Mean ± SD T1 T2 Stand-to-sit ROM of the knee joint angles (degrees) Left knee 77.70 ± 11.63 79.12 ± 11.24 Right knee 79.66 ± 10.91 80.30 ± 12.90 Sit-to-stand ROM of the knee joint angles (degrees) Left knee 71.66 ± 10.46 70.22 ± 9.49 Right knee 74.18 ± 10.88 72.89 ± 11.44 Walking ROM of the knee joint angles (degrees) Left knee 52.60 ± 3.50 54.51 ± 5.02 Right knee 52.67 ± 3.16 54.89 ± 8.65 Spatiotemporal gair parameters Stride length (cm) 73.62 ± 7.90 78.98 ± 13.12 Stride time (s) 1.43 ± 0.16 1.38 ± 0.16 Table 6.11 presents the virtual challenge scores for each trial of the virtual challenge Watch Tv. 76 Table 6.11 – Television state achieved in each trial of the virtual challenge “Watch Tv” Participants Television state achieved T1 T2 P1 On On P2 On On P3 On On P4 On On P5 On On P6 On On Concerning the virtual challenge “Watch Tv”, none of the stand-to-sit and sit-to-stand motor tasks presented statistically significant differences between trials once the p -value is higher than 0.05 (Table 6.9). For these two motor tasks, improvements were found for COM displacement in the ML axis (DispML_Stand), and maximum COM velocity in AP (VelMaxAP_Stand) and ML (VelMaxML_Stand) directions. For these three-assessment metrics, in general, participants showed an increase in balance in the ML axis (with decreasing values from T1 to T2), an increase in velocity in forward movements, and a decrease in velocity in lateral movements, while performing the task from sitting to standing. Regarding the walking motor task, two variables (DispML_Walk, and VelMinAP_Walk) showed statistically significant differences between first and second trials ( p -value = 0.049, and p -value = 0.020, respectively) (Table 6.9). These results suggest a significant improvement in balance in the ML direction since the COM displacement in this axis decreased considerably from T1 (9.113 ± 0.982 cm) to T2 (7.483 ± 0.553 cm), and that participants were able to walk significantly faster in the direction of movement in the second trial (46.773 ± 3.836 cm/s) compared to the first one (42.647 ± 3.259 cm/s) since the minimum COM velocity in the AP axis increased. The decreasing of the stride time (Table 6.10) reinforces this statement. Although the other variables for this motor task did not reveal statistically significant differences between trials, all of them presented an improvement in balance function. COM displacement in the AP direction (DispAP_Walk) increased between trials, meaning that in the second trial participants were able to perform a greater COM displacement in this direction (77.713 ± 5.760 cm), which indicates a greater distance achieved during a gait cycle compared to the first trial (73.708 ± 3.391) (Table 6.9). This explanation can be supported by the increase of the stride length values, showed in Table 6.10. The maximum COM velocity in AP axis, presenting positive values, also increased (Table 6.9), complementing the explanation of the participants’ faster velocity in this direction. The COM velocity 77 for left (VelMaxML_Walk) and right (VelMinML_Walk) sides of ML direction decreased from T1 to T2 (Table 6.9), meaning more stable movements to the sides while walking forwards. All the participants were able to turn on the television (Table 6.11). None of the studies reviewed in the literature assesses the user’s performance during the VR training. Thus, it is impossible to perform a comparison between the VR training data from the present study with those from the literature. 6.2 Before vs after VR intervention From data acquisition trials carried out before and after VR intervention, primary and secondary outcome measures for the BBS and TUG clinical tests were computed (Table 0.9 to Table 0.14, Appendix I). These outcomes were statistically evaluated, comparing the primary outcomes from before versus after VR intervention for each BBS’s and TUG’s task. Tables 6.12 and 6.13 present the results of parametric (paired samples t -test) and nonparametric metrics (Wilcoxon signed-rank test) test, for BBS’s and TUG’s tasks, respectively. 78 Table 6.12 - Mean and SD of pre and post VR training trials, and the p-value of the paired samples test and Wilcoxon test for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for BBS’ tasks, considering the time condition and a significance level of 5 %. The most representative results ( p -value < 0.05) appear in bold Paired Samples Test Test Statistics a BBS Mean ± SD t -test Wilcoxon Pre Post t Sig. (2tailed) Z Asymp. Sig. (2tailed) Sit-to-stand DispAP_BBS1 42.235 ± 1.445 37.548 ± 2.344 2.629 .047 - - DispML_BBS1 2.573 ± 0.229 2.550 ± 0.426 .049 .962 - - VelMinAP_BBS1 -6.917 ± 1.502 -10.002 ± 4.041 - - -.105c .917 VelMaxAP_BBS1 48.117 ± 2.860 38.878 ± 8.081 1.444 .208 - - VelMinML_BBS1 -3.913 ± 0.631 -3.523 ± 0.680 -1.088 .326 - - VelMaxML_BBS1 2.910 ± 0.819 4.030 ± 0.800 -1.477 .200 - - Standing unsupported DispAP_BBS2 2.200 ± 0.312 2.362 ± 0.510 - - -.524c .600 DispML_BBS2 0.822 ± 0.117 0.912 ± 0.111 -.519 .626 - - VelMinAP_BBS2 -1.990 ± 0.627 -2.203 ± 0.706 - - -.524b .600 VelMaxAP_BBS2 1.108 ± 0.152 1.137 ± 0.255 -.155 .883 - - VelMinML_BBS2 -1.057 ± 0.496 -1.440 ± 0.522 - - -1.156b .248 VelMaxML_BBS2 0.705 ± 1.131 0.642 ± 0.056 .423 .690 - - Sitting unsupported DispAP_BBS3 1.647 ± 0.408 1.382 ± 0.111 - - -.314c .753 DispML_BBS3 1.118 ± 0.347 0.705 ± 0.162 1.122 .313 - - VelMinAP_BBS3 -5.513 ± 1.652 -5.135 ± 1.399 -.924 .398 - - VelMaxAP_BBS3 6.375 ± 2.613 6.880 ± 3.237 -.305 .773 - - VelMinML_BBS3 -3.167 ± 0.625 -2.240 ± 0.511 -2.096 .090 - - VelMaxML_BBS3 1.303 ± 0.644 2.335 ± 0.907 - - -1.782b .075 79 Paired Samples Test Test Statistics a BBS Mean ± SD t -test Wilcoxon Pre Post t Sig. (2tailed) Z Asymp. Sig. (2tailed) Stand-to-sit DispAP_BBS4 40.058 ± 1.844 35.810 ± 2.045 3.969 .011 - - DispML_BBS4 3.950 ± 0.898 3.067 ± 0.685 1.659 .158 - - VelMinAP_BBS4 -39.522 ± 4.125 -38.225 ± 4.337 -.351 .740 - - VelMaxAP_BBS4 2.472 ± 0.453 3.663 ± 0.631 -2.729 .041 - - VelMinML_BBS4 -5.257 ± 1.570 -4.520 ± 1.064 -.559 .600 - - VelMaxML_BBS4 2.307 ± 0.363 2.413 ± 0.331 -.178 .866 - - Transfers DispAP_BBS5 95.915 ± 3.175 95.400 ± 4.164 .168 .874 - - DispML_BBS5 13.653 ± 1.605 12.683 ± 0.964 .565 .597 - - VelMinAP_BBS5 -38.720 ± 2.442 -41.933 ± 2.678 3.004 .030 - - VelMaxAP_BBS5 43.570 ± 3.222 46.175 ± 4.696 -.846 .436 - - VelMinML_BBS5 -17.305 ± 0.811 -19.795 ± 1.562 2.148 .084 - - VelMaxML_BBS5 19.380 ± 0.487 18.297 ± 2.408 - - -.105c .917 Standing with eyes closed DispAP_BBS6 1.390 ± 0.244 1.435 ± 0.209 -.182 .862 - - DispML_BBS6 0.513 ± 0.094 0.685 ± 0.147 -1.478 .199 - - VelMinAP_BBS6 -2.017 ± 0.610 -1.635 ± 0.576 -3.452 .018 - - VelMaxAP_BBS6 1.090 ± 0.179 1.270 ± 1.184 -3.286 .022 - - VelMinML_BBS6 -1.190 ± 0.585 -0.905 ± 0.252 - - -.734b .463 VelMaxML_BBS6 0.557 ± 0.089 0.673 ± 0.151 -1.159 .299 - - 80 Paired Samples Test Test Statistics a BBS Mean ± SD t -test Wilcoxon Pre Post t Sig. (2tailed) Z Asymp. Sig. (2tailed) Standing with feet together DispAP_BBS7 2.010 ± 0.330 2.275 ± 0.491 -.804 .458 - - DispML_BBS7 1.313 ± 0.093 1.565 ± 0.145 - - -1.992c .046 VelMinAP_BBS7 -2.100 ± 0.732 -2.583 ± 0.648 .763 .480 - - VelMaxAP_BBS7 1.093 ± 0.165 1.162 ± 0.154 -.465 .662 - - VelMinML_BBS7 -1.088 ± 0.177 -1.187 ± 0.233 .462 .663 - - VelMaxML_BBS7 1.143 ± 0.174 1.792 ± 0.517 - - -1.992c .046 Reaching forward DispAP_BBS8 9.773 ± 0.949 8.365 ± 0.934 6.890 .001 - - DispML_BBS8 3.200 ± 0.533 3.178 ± 0.474 .026 .980 - - VelMinAP_BBS8 -11.645 ± 0.956 -8.870 ± 1.391 -2.042 .097 - - VelMaxAP_BBS8 7.077 ± 1.128 8.023 ± 1.009 -1.395 .222 - - VelMinML_BBS8 -2.770 ± 0.335 -3.337 ± 0.647 - - -.734b .463 VelMaxML_BBS8 3.853 ± 0.597 3.743 ± 0.875 .082 .938 - - Pick up an object DispAP_BBS9 6.600 ± 0.928 5.560 ± 0.832 1.370 .229 - - DispML_BBS9 3.762 ± 0.978 3.380 ± 1.101 .815 .452 - - VelMinAP_BBS9 -7.602 ± 0.888 -6.563 ± 1.324 -.598 .576 - - VelMaxAP_BBS9 7.013 ± 0.773 6.950 ± 1.271 .075 .943 - - VelMinML_BBS9 -4.057 ± 0.670 -4.000 ± 0.873 -.140 .894 - - VelMaxML_BBS9 4.240 ± 0.861 4.797 ± 0.761 -.694 .519 - - 81 Paired Samples Test Test Statistics a Mean ± SD t -test Wilcoxon BBS Pre Post t Sig. (2tailed) Z Asymp. Sig. (2tailed) Looking over shoulders DispAP_BBS10 3.677 ± 0.438 3.717 ± 0.215 -.139 .895 - - DispML_BBS10 2.335 ± 0.324 2.912 ± 0.395 -1.121 .313 - - VelMinAP_BBS10 -3.200 ± 0.667 -3.727 ± 0.539 - - -1.363b .173 VelMaxAP_BBS10 3.208 ± 0.486 3.780 ± 0.539 -1.445 .208 - - VelMinML_BBS10 -3.018 ± 0.389 -3.132 ± 0.377 .189 .858 - - VelMaxML_BBS10 2.458 ± 0.360 4.037 ± 0.552 -2.526 .053 - - Turn 360 degrees DispAP_BBS11 21.392 ± 2.707 20.437 ± 2.482 .265 .801 - - DispML_BBS11 22.605 ± 1.610 21.595 ± 1.820 .418 .694 - - VelMinAP_BBS11 -27.698 ± 3.648 -29.243 ± 2.805 .363 .731 - - VelMaxAP_BBS11 26.020 ± 3.104 23.132 ± 1.501 .807 .457 - - VelMinML_BBS11 -25.968 ± 2.229 -22.338 ± 1.452 -1.111 .317 - - VelMaxML_BBS11 32.145 ± 3.925 25.462 ± 1.605 2.064 .094 - - Place alternate foot on step DispAP_BBS12 6.770 ± 0.542 7.402 ± 0.836 -.530 .619 - DispML_BBS12 16.312 ± 0.772 15.590 ± 0.129 - - -.314b .753 VelMinAP_BBS12 -8.037 ± 0.368 -7.393 ± 0.655 -1.264 .262 - - VellMaxAP_BBS12 9.703 ± 0.834 9.185 ± 0.907 .791 .465 - - VelMinML_BBS12 -19.775 ± 1.755 -19.087 ± 0.800 - - -.315c .752 VelMaxML_BBS12 20.133 ± 1.261 19.910 ± 0.772 .148 .888 - - 82 Paired Samples Test Test Statistics a BBS Mean ± SD t -test Wilcoxon Pre Post t Sig. (2tailed) Z Asymp. Sig. (2tailed) Standing with one foot in front DispAP_BBS13 1.630 ± 0.333 1.745 ± 0.173 -.291 .783 - - DispML_BBS13 1.943 ± 0.343 1.780 ± 0.098 - - -.943b .345 VelMinAP_BBS13 -2.052 ± 0.622 -2.342 ± 0.613 - - -.524b .600 VelMaxAP_BBS13 1.730 ± 0.551 2.140 ± 0.758 - - -.943c .345 VelMinML_BBS13 -1.418 ± 0.114 -1.667 ± 0.296 .998 .364 - - VelMaxML_BBS13 1.888 ± 0.265 2.535 ± 0.735 -.932 .394 - - Standing on one leg DispAP_BBS14 4.508 ± 0.707 4.962 ± 0.805 - - -.524c .600 DispML_BBS14 9.367 ± 1.690 9.083 ± 1.068 .262 .804 - - VelMinAP_BBS14 -3.853 ± 0.670 -2.853 ± 0.471 -2.943 .032 - - VelMaxAP_BBS14 6.287 ± 1.656 7.005 ± 0.916 -.376 .723 - - VelMinML_BBS14 -9.307 ± 2.760 -8.467 ± 2.769 -.803 .459 - - VelMaxML_BBS14 9.520 ± 2.579 7.417 ± 1.974 .771 .476 - - a. Wilcoxon Signed Ranks Test b. Based on positive ranks c. Based on negative ranks 83 Table 6.13 - Mean and SD of T1 and T2, and the p-value of the paired samples test and Wilcoxon test for COM displacement (Disp) and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for sit-to-stand, walking, and stand-to-sit motor tasks of TUG, considering the time condition and a significance level of 5 %. The most representative results ( p -value < 0.05) appear in bold Paired Samples Test Test Statistics a TUG Mean ± SD t -test Wilcoxon Pre Post t Sig. (2tailed) Z Asymp. Sig. (2tailed) Sit-to-stand DispAP_Stand 38.928 ± 3.046 37.310 ± 3.122 .782 .470 - - DispML_Stand 5.207 ± 0.711 5.227 ± 0.842 -.021 .984 - - VelMinAP_Stand -5.222 ± 1.704 -6.020 ± 1.714 .802 .459 - - VelMaxAP_Stand 50.185 ± 4.329 47.108 ± 3.848 .960 .381 - - VelMinML_Stand -6.865 ± 1.908 -8.640 ± 4.153 - - -.105c .917 VelMaxML_Stand 10.277 ± 3.422 9.408 ± 2.309 .247 .815 - - Walking DispAP_Walk 115.265 ± 2.603 113.800 ± 2.035 1.147 .303 - - DispML_Walk 15.330 ± 2.321 15.270 ± 2.271 .025 .981 - - VelMinAP_Walk 85.905 ± 3.447 84.233 ± 3.741 1.479 .199 - - VelMaxAP_Walk 120.197 ± 3.172 113.738 ± 3.866 2.617 .047 - - VelMinML_Walk -18.453 ± 3.470 -22.898 ± 4.488 1.104 .320 - - VelMaxML_Walk 17.775 ± 4.525 11.605 ± 4.601 - - -1.572b .116 Stand-to-sit DispAP_Sit 34.670 ± 2.331 36.375 ± 3.047 -1.159 .299 - - DispML_Sit 5.095 ± 1.144 5.963 ± 0.908 -.622 .561 - - VelMinAP_Sit -35.328 ± 2.764 -36.767 ± 2.955 1.476 .200 - - VelMaxAP_Sit 1.813 ± 0.266 1.622 ± 0.347 .554 .604 - - VelMinML_Sit -11.137 ± 3.630 -12.022 ± 2.256 .164 .876 - - VelMaxML_Sit 5.520 ± 2.403 5.368 ± 1.882 - - -.524c .600 a. Wilcoxon Signed Ranks Test b. Based on positive ranks c. Based on negative ranks 90 them to understand their limitations in posture and balance (P6), as well as to increase balance and trunk mobility (P5), and also to recover from physical injuries of the lower body (P4). Despite the high Effort/Importance experienced by the participants, they reported feeling no Pressure/Tension (3.03 of mean) while performing the virtual challenges. Table 6.19 - Mean and standard deviation (SD) of all the participants, for each subscale of IMI IMI Interest/ Enjoyment Perceived Competence Effort/ Importance Pressure/ Tension Value/ Usefulness Mean ± SD 5.81 ± 0.61 5.53 ± 0.58 5.27 ± 0.98 3.03 ± 1.53 5.33 ± 0.84 Best score 7.00 7.00 2.20 1.00 7.00 Table 6.20 - Participants’ answers to the open questions of the Value/Usefulness subscale of IMI questionnaire Open-ended questions of Value/Usefulness subscale “I think that doing this activity is useful for…” “I think is important to do because it can…” “I think doing this activity could help me to…” P1 “improving my balance” “improving my balance” “improving my balance” P2 “practice certain motor skills on injured patients, or patients suffering from some illness that restricts their movement” “help recover motor skills in controlled environments” “recover from physical injuries to the lower body” “people with movements constraints” “develop specific movements” “exercise, practise and develop some movements that are hard for me to do” P3 P4 “retrain people after some accident or illness” “help people with motor difficulties” “improve my equilibrium and stability” “training balance and mobility” “prompt self-awareness about balance, mobility and maybe ways to improve it” “improve balance, focus and trunk mobility” P5 P6 “evaluate the progress of people with postural and stability deficits.” “improve the stability, postural, and balance performance of people with balance losses.” “understand my postural and balance limits.” 91 6.4 Conclusions During VR intervention, VelMaxAP_Cook ( p -value = 0.027), DispML_Walk ( p -value = 0.049), and VelMinAP_Walk ( p -value = 0.020) are the only variables presenting statistically significant differences between the first and second trial of VR intervention. The lack of more significant differences between trials may be because the participants were all healthy and the reduced number of trials and VR sessions. By comparing the balance performance before and after the VR intervention, statistically significant differences were found for the following BBS’s tasks: DispAP_BBS1 ( p -value = 0.047), DispAP_BBS4 ( p - value = 0.011), VelMaxAP_BBS4 ( p -value = 0.041), VelMinAP_BBS5 ( p -value = 0.030), VelMinAP_BBS6 ( p -value = 0.018), VelMaxAP_BBS6 ( p -value = 0.022), DispML_BBS7 ( p -value = 0.046), VelMaxML_BBS7 ( p -value = 0.046), DispAP_BBS8 ( p -value = 0.001), and VelMinAP_BBS14 ( p -value = 0.032) and for TUG’s task: VelMaxAP_Walk ( p -value = 0.047). Clinical scales scores showed a decreased time to perform the TUG clinical scale, as expected. The BBS scores (before: 55.83 points, after: 55.83 points) are in accordance with those expected for healthy people (values between 41 and 56 points) [78], with the TUG scores (before: 10.67 s, after: 10.48 s) at the threshold of the same classification (less than 10 s) [79]. Sensor-based measures enable to find significant improvements during clinical test whereas no significant differences were found in the BBS and TUG scores. This shows the relevance for collecting objective metrics even when performing clinical tests. Overall, the proposed VR-based balance training tool was able to improve stability in virtual challenges performed by healthy subjects, mainly at the walking level, with only two trials. Besides the balance results being encouraging, the tool proved to be well accepted by the users. User’s experience evaluation tests proved the high acceptability of the VR-based tool by the participants, who were interested during the VR intervention, experiencing a sense of presence, involvement, and realism by the VRE created. Findings point out that the tool seems promising for the rehabilitation of static and dynamic balance as a complement to conventional physical therapy. 92 7. CONCLUSIONS Stroke has a negative impact worldwide, being the leading cause of long-term disability, usually including loss of balance that limits the survivor’s daily life. However, the survivors may regain their motor independence through neuroplasticity phenomenon, achieved by rehabilitation, including rehabilitation driven by robotic devices, such as VR tools. There is evidence that using VR tools as a complement to physical therapies improves and accelerates the user’s recovery, as VREs can be customised to the user’s imminent needs and provide intensive, repetitive, and enthusiastic training. In this sense, this dissertation designed, developed, and preliminarily validated a fully immersive VR tool, based on a user-centered design, to maximize the user’s recovery, acceptability, and enthusiasm. The state of the art on VR tools developed for post-stroke balance rehabilitation reveals that there is a lack of VR tools to be used in this field with a user-centered design. From current studies, there is a prevalence to use screens as VR technology, followed by HMD. However, there is a lack of combination of these two technologies. Cameras and balance boards are the most used sensors. The motor tasks most performed are weight shifting/bearing, treadmill walking, stepping, pelvic/hip and knee movements. Most of the VR tools comprise a closed-loop control, with virtual challenges generally aiming to catch, avoid, and navigate virtual objects, usually belonging to commercial systems with the main purpose of entertainment, with no user-centered design solutions. However, walking was mainly addressed alone, and in an open loop, in opposition to the remaining motor tasks. Thus, in this sense, there is space to create wearable VR-based tools combining walking with other motor tasks and in a closed-loop. Visual and auditory feedback are the stimuli mostly used, having a lack of vibrotactile actuators in VR tools. BBS and TUG are the most used clinical scales to assess static and dynamic balance that are the main difficulties in post-stroke patients. The designed VR-based tool has a fully wearable design. It combines the HTC Vive Pro to immerse the user in the VRE through HMD, and the Xsens MVN Awinda, in a full-body configuration, as well as the HTC controllers to allow the user to interact with the VRE. The developed tool provides visual, sonorous, and vibrotactile stimulation through HMD, built-in headphones, and HTC controllers, respectively. In addition to the HMD, a screen was used to allow the therapist to follow, in real-time, the user’s performance during balance training and, consequently, provide additional support to patients through auditory instructions. The developed VR tool is based on the user’s home ADLs and attends to the most performed motor tasks in balance training-related literature and in the BBS and TUG clinical tests. The VR tool addresses nine motor tasks such as walking, stepping, sit-to-stand and stand-to-sit, knee flexion and extension, 93 weight shifting, trunk and pelvic/hip movement, looking over the shoulders, standing upright on one leg, and reach with hands as far as possible. They were included in four virtual challenges, namely, Fruit Catcher, Cooking, Take a Shower, and Watch TV. The integration of first and second-person views of an avatar mimicking user’s movements in real-time, grab objects, controllers’ vibration, real-time COM visualization in the second-person view, turn on the Tv, shower and bathtub rim’s sounds, stepping up the tree ladder, and customise the kitchen shelves’ height, were the main control strategies develop to create dynamism to the static VRE, and allow the user to interact and receive real-time feedback. A questionnaire on the ADLs most performed by people and that require balance was developed as a parallel study. The results reveal that the activities most performed and appreciated by the general population were in line with three of the four virtual challenges: Cooking, Take a Shower, and Watch Tv. Furthermore, it was observed that age influences the most appreciated ADLs, being a factor that should be considered in future research when developing virtual challenges. The validation protocol was carried out, including six healthy and young participants with any type of balance injuries. Healthy subjects were enrolled since this preliminary validation aims to assess the user’s experience and operability of the proposed VR tool, and assess the impact (either positive or negative) on motor function. All participants felt well throughout the VR sessions, that includes rest breaks, with no negative occurrences to report. Primary and secondary sensor-based outcomes were successfully measured through Xsens MVN Awinda before, during, and after VR intervention. In addition, virtual challenges scores were considered, as well the user’s experience evaluation tests IMI and IPQ, aiming to assess the user’s subjective experienced related to the performed activity, and the sense of presence in the VRE. The results from the validation protocol showed statistically significant differences in COM displacement in ML direction, and minimum COM velocity in AP direction (Watch Tv, mainly at the walking level), and maximum COM velocity in AP direction (Cooking virtual challenge) between consecutive trials of the VR training. Although significantly, the latter shows a decreased velocity, which is not desired in balance training. Virtual challenges scores showed improvements for some of the participants in the number of apples caught (Fruit Catcher virtual challenge), in the number of times the user kicked the bathtub and in the time the sound played inside the bathtub (Take a Shower virtual challenge). All other virtual challenges scores presented the maximum score for all participants. Further improvements were reported, by comparing the balance function before and after VR intervention. TUG improved the clinical assessment by decreasing the time to perform the whole test, and BBS maintained the results since all participants, except one, presented the maximum score for both trials before and after VR training due to 94 their healthy condition. These findings reveal that this tool can be effective for balance rehabilitation, as a complement to conventional physical therapies. The results from IMI’s questionnaire proved that participants felt a higher interest and enjoyment, as well as a higher competence while performing the virtual challenges, proving the high acceptability of the tool. Moreover, participants revealed that they see value and usefulness in the tool for improving not only balance and posture, but also lower limb injuries. The IPQ questionnaire revealed that participants felt immersed in the virtual world due to the level of realism and sense of presence experienced, which is important for the best and fast improvements. This dissertation allows to answer the RQs appointed in Chapter 1: • RQ1: What specifications should be considered for the user-centered design of VR-based post-stroke balance rehabilitation? Chapter 2 answered this RQ. VR technology, sensor integration, motor tasks, and virtual challenges, as well as the type of feedback received by the participants are the specifications that should be considered for the design on a VR tool for balance rehabilitation. There is a need for using wearable VR and sensor technologies to enable balance training during static and dynamic motor tasks. The virtual challenges should encourage the users to perform daily motor tasks, which should be selected in accordance with the user’s needs in ADLs and preferences towards a user-centered design. Moreover, it is necessary to ensure that the VR enables multimodal feedback (visual, auditory, and haptic stimuli), commonly received in daily living, to allow a more realistic experience. • RQ2: What factors should be considered for a user-centered design of a VR-based tool in balance training? Chapter 4 answered this RQ. According to the participants’ answers to the questionnaire about ADLs that require balance, among the assessed factors, age showed to influence the most appreciated ADLs. Thus, future research should consider age when designing virtual challenges, to create a more user-centered design VR tool, since the activities most enjoyed by people differ with aging. • RQ3: Which are the effects of a fully immersive VR-based tool in balance training? Chapter 6 answered this RQ. The primary outcomes revealed a statistically significant improvement of the minimum COM velocity in AP direction ( p -value = 0.020) and in the COM 95 displacement in the ML direction ( p -value = 0.049) in the walking motor task of the virtual challenge Watch Tv. This suggests that, from the first to the second trial, participants were able not only to walk faster but to do so with more stable balance control, decreasing their side-to-side movement directions. In addition to these metrics, many others showed considerable improvements, although not enough to be significant. The results provided by the clinical tests BBS and TUG also revealed appreciable improvements in some of the metrics assessed. Furthermore, the total time to perform the TUG decreased with only two trials of VR training (before VR training: 10.673 ±0.525 s; after VR training: 10.480 ± 0.437 s). 7.1 Future Work The future work comprises the following research directions: (i) to increase the number of virtual challenges, attending to the ADLs reported in the questionnaire, namely driving a car and dancing; (ii) to increase the number of motor tasks addressed per virtual challenge; (iii) to implement different levels of difficulty in each virtual challenge so that it can be used by patients with different disability levels and progressively increase the level of difficulty as they improve; (iv) to customise all virtual challenges according to user’s anthropometric data; (v) to update the virtual challenges of the VR-based tool considering the age factor (vi) to carry out a more balanced study of the factors that influence the most performed and appreciated ADLs; (vii) to improve the aesthetic design of the VR tool by creating a more interactive menu with instructions and a window on the screen with the user’s real time score, according to the performance in the virtual challenge; (viii) to perform a validation protocol with end-users with balance disorders (e.g., post-stroke patients, elderly people) through a randomised controlled study with a longer training by increasing the number of VR sessions. Further work involves scientific dissemination of the achieved results in peer-review ISI/Scopus journals and communications activities next to hospitals, rehabilitation clinics, and care facilities to motivate their interest to join in the end-user’s validation study. 96 REFERENCES [1] “WHO EMRO | Stroke, Cerebrovascular accident | Health topics.” http://www.emro.who.int/health-topics/stroke-cerebrovascular-accident/index.html (accessed Jan. 21, 2021). [2] R. Gupta, “Stroke: a leading cause of long term disability.” https://www.fortishealthcare.com/blog/strokeleading-long-term-disability/ (accessed Oct. 31, 2021). [3] “Balance problems - Symptoms and causes - Mayo Clinic.” https://www.mayoclinic.org/diseasesconditions/balance-problems/symptoms-causes/syc-20350474 (accessed Oct. 30, 2021). [4] “Balance Disorders — Causes, Types & Treatment | NIDCD.” https://www.nidcd.nih.gov/health/balance-disorders (accessed Oct. 30, 2021). [5] “Balance | Definition of Balance by Merriam-Webster.” https://www.merriamwebster.com/dictionary/balance (accessed Oct. 31, 2021). [6] “Balance - Harvard Health.” https://www.health.harvard.edu/topics/balance (accessed Oct. 31, 2021). [7] “Falls.” https://www.who.int/news-room/fact-sheets/detail/falls (accessed Oct. 31, 2021). [8] “Basic Facts about Basic Facts | Aging & Health A-Z | American Geriatrics Society | HealthInAging.org.” https://www.healthinaging.org/a-z-topic/balance-problems/basic-facts (accessed Oct. 31, 2021). [9] M. H. Kaveh, F. Bahadori, A. Doosti, and A. Asadollahi, “The effect of balance exercise training on balance status, and quality of life in elderly women: A randomized controlled trial,” Malaysian J. Med. Heal. Sci. , vol. 17, no. 2, pp. 129–136, 2021. [10] Courtney E. Ackerman, “What is Neuroplasticity? A Psychologist Explains [+14 Exercises].” https://positivepsychology.com/neuroplasticity/ (accessed Oct. 31, 2021). [11] U. B. Rahayu, S. Wibowo, I. Setyopranoto, and M. Hibatullah Romli, “Effectiveness of physiotherapy interventions in brain plasticity, balance and functional ability in stroke survivors: A randomized controlled trial,” NeuroRehabilitation , vol. 47, no. 4, pp. 463–470, 2020, doi: 10.3233/NRE-203210. [12] C. Rossano and P. Terrier, “Visually-guided gait training in paretic patients during the first rehabilitation phase: Study protocol for a randomized controlled trial,” Trials , vol. 17, no. 1, pp. 1–9, 2016, doi: 10.1186/s13063-016-1630-8. [13] S. Guo, H. Cai, and J. Guo, “A Novel Fuzzy Neural Network-based Rehabilitation Stage Classifying 97 Method for the Upper Limb Rehabilitation Robotic System,” 2020 IEEE Int. Conf. Mechatronics Autom. ICMA 2020 , pp. 261–266, 2020, doi: 10.1109/ICMA49215.2020.9233519. [14] V. Klamroth-Marganska, “Stroke rehabilitation: Therapy robots and assistive devices,” Adv. Exp. Med. Biol. , vol. 1065, pp. 579–587, 2018, doi: 10.1007/978-3-319-77932-4_35. [15] R. Gassert and V. Dietz, “Rehabilitation robots for the treatment of sensorimotor deficits: A neurophysiological perspective,” J. Neuroeng. Rehabil. , vol. 15, no. 1, pp. 1–15, 2018, doi: 10.1186/s12984-018-0383-x. [16] T. Bao et al. , “Effects of long-term balance training with vibrotactile sensory augmentation among community-dwelling healthy older adults: a randomized preliminary study,” J. Neuroeng. Rehabil. , vol. 15, no. 1, p. 5, Dec. 2018, doi: 10.1186/s12984-017-0339-6. [17] G. Tieri, G. Morone, S. Paolucci, and M. Iosa, “Virtual reality in cognitive and motor rehabilitation: facts, fiction and fallacies,” Expert Rev. Med. Devices , vol. 15, no. 2, pp. 107–117, 2018, doi: 10.1080/17434440.2018.1425613. [18] F. Arcuri, C. Porcaro, I. Ciancarelli, P. Tonin, and A. Cerasa, “Electrophysiological correlates of virtual-reality applications in the rehabilitation setting: New perspectives for stroke patients,” Electron. , vol. 10, no. 7, 2021, doi: 10.3390/electronics10070836. [19] M. C. Howard, “A meta-analysis and systematic literature review of virtual reality rehabilitation programs,” Comput. Human Behav. , vol. 70, pp. 317–327, 2017, doi: 10.1016/j.chb.2017.01.013. [20] M. Francesca, “Virtual Reality Rehabilitation - The Future of Physical Therapy.” https://imotions.com/blog/virtual-reality-rehabilitation/ (accessed Nov. 01, 2021). [21] A. Asadzadeh, T. Samad-Soltani, Z. Salahzadeh, and P. Rezaei-Hachesu, “Effectiveness of virtual reality-based exercise therapy in rehabilitation: A scoping review,” Informatics Med. Unlocked , vol. 24, p. 100562, 2021, doi: 10.1016/j.imu.2021.100562. [22] R. Mohammadi, A. V. Semnani, M. Mirmohammadkhani, and N. Grampurohit, “Effects of Virtual Reality Compared to Conventional Therapy on Balance Poststroke: A Systematic Review and MetaAnalysis,” J. Stroke Cerebrovasc. Dis. , vol. 28, no. 7, pp. 1787–1798, Jul. 2019, doi: 10.1016/j.jstrokecerebrovasdis.2019.03.054. [23] D. Corbetta, F. Imeri, and R. Gatti, “Rehabilitation that incorporates virtual reality is more effective than standard rehabilitation for improving walking speed, balance and mobility after stroke: a systematic review,” J. Physiother. , vol. 61, no. 3, pp. 117–124, Jul. 2015, doi: 10.1016/j.jphys.2015.05.017. 98 [24] L. Barcala, L. A. C. Grecco, F. Colella, P. R. G. Lucareli, A. S. I. Salgado, and C. S. Oliveira, “Visual Biofeedback Balance Training Using Wii Fit after Stroke: A Randomized Controlled Trial,” J. Phys. Ther. Sci. , vol. 25, no. 8, pp. 1027–1032, 2013, doi: 10.1589/jpts.25.1027. [25] K. H. Cho, K. J. Lee, and C. H. Song, “Virtual-Reality Balance Training with a Video-Game System Improves Dynamic Balance in Chronic Stroke Patients,” Tohoku J. Exp. Med. , vol. 228, no. 1, pp. 69–74, 2012, doi: 10.1620/tjem.228.69. [26] K. H. Cho and W. H. Lee, “Virtual Walking Training Program Using a Real-world Video Recording for Patients with Chronic Stroke,” Am. J. Phys. Med. Rehabil. , vol. 92, no. 5, pp. 371–384, May 2013, doi: 10.1097/PHM.0b013e31828cd5d3. [27] K. H. Cho and W. H. Lee, “Effect of treadmill training based real-world video recording on balance and gait in chronic stroke patients: A randomized controlled trial,” Gait Posture , vol. 39, no. 1, pp. 523–528, Jan. 2014, doi: 10.1016/j.gaitpost.2013.09.003. [28] J. S. Huh, Y.-S. Lee, C.-H. Kim, Y.-S. Min, M.-G. Kang, and T.-D. Jung, “Effects of Balance Control Training on Functional Outcomes in Subacute Hemiparetic Stroke Patients,” Ann. Rehabil. Med. , vol. 39, no. 6, p. 995, 2015, doi: 10.5535/arm.2015.39.6.995. [29] T. In, K. Lee, and C. Song, “Virtual Reality Reflection Therapy Improves Balance and Gait in Patients with Chronic Stroke: Randomized Controlled Trials,” Med. Sci. Monit. , vol. 22, pp. 4046– 4053, Oct. 2016, doi: 10.12659/MSM.898157. [30] J. H. Kim, S. H. Jang, C. S. Kim, J. H. Jung, and J. H. You, “Use of Virtual Reality to Enhance Balance and Ambulation in Chronic Stroke: A Double-Blind, Randomized Controlled Study,” Am. J. Phys. Med. Rehabil. , vol. 88, no. 9, pp. 693–701, Sep. 2009, doi: 10.1097/PHM.0b013e3181b33350. [31] S. H. Lee et al. , “Feasibility and Effects of Newly Developed Balance Control Trainer for Mobility and Balance in Chronic Stroke Patients: A Randomized Controlled Trial,” Ann. Rehabil. Med. , vol. 36, no. 4, p. 521, 2012, doi: 10.5535/arm.2012.36.4.521. [32] I.-W. Lee, Y.-N. Kim, and D.-K. Lee, “Effect of a virtual reality exercise program accompanied by cognitive tasks on the balance and gait of stroke patients,” J. Phys. Ther. Sci. , vol. 27, no. 7, pp. 2175–2177, 2015, doi: 10.1589/jpts.27.2175. [33] R. Lloréns, J.-A. Gil-Gómez, M. Alcañiz, C. Colomer, and E. Noé, “Improvement in balance using a virtual reality-based stepping exercise: a randomized controlled trial involving individuals with chronic stroke,” Clin. Rehabil. , vol. 29, no. 3, pp. 261–268, Mar. 2015, doi: 10.1177/0269215514543333. 99 [34] Y.-H. Park, C. Lee, and B.-H. Lee, “Clinical usefulness of the virtual reality-based postural control training on the gait ability in patients with stroke,” J. Exerc. Rehabil. , vol. 9, no. 5, pp. 489–494, Oct. 2013, doi: 10.12965/jer.130066. [35] D.-S. Park, D.-G. Lee, K. Lee, and G. Lee, “Effects of Virtual Reality Training using Xbox Kinect on Motor Function in Stroke Survivors: A Preliminary Study,” J. Stroke Cerebrovasc. Dis. , vol. 26, no. 10, pp. 2313–2319, Oct. 2017, doi: 10.1016/j.jstrokecerebrovasdis.2017.05.019. [36] B. S. Rajaratnam et al. , “Does the Inclusion of Virtual Reality Games within Conventional Rehabilitation Enhance Balance Retraining after a Recent Episode of Stroke?,” Rehabil. Res. Pract. , vol. 2013, pp. 1–6, 2013, doi: 10.1155/2013/649561. [37] Y. B. Song, M. H. Chun, W. Kim, S. J. Lee, J. H. Yi, and D. H. Park, “The Effect of Virtual Reality and Tetra-Ataxiometric Posturography Programs on Stroke Patients With Impaired Standing Balance,” Ann. Rehabil. Med. , vol. 38, no. 2, p. 160, 2014, doi: 10.5535/arm.2014.38.2.160. [38] G. I. Yatar and S. A. Yildirim, “Wii Fit balance training or progressive balance training in patients with chronic stroke: a randomised controlled trial,” J. Phys. Ther. Sci. , vol. 27, no. 4, pp. 1145– 1151, 2015, doi: 10.1589/jpts.27.1145. [39] D. L. Jaffe, D. A. Brown, C. D. Pierson-Carey, E. L. Buckley, and H. L. Lew, “Stepping over obstacles to improve walking in individuals with poststroke hemiplegia,” J. Rehabil. Res. Dev. , vol. 41, no. 3A, p. 283, 2004, doi: 10.1682/JRRD.2004.03.0283. [40] J. Jung, J. Yu, and H. Kang, “Effects of Virtual Reality Treadmill Training on Balance and Balance Self-efficacy in Stroke Patients with a History of Falling,” J. Phys. Ther. Sci. , vol. 24, no. 11, pp. 1133–1136, 2012, doi: 10.1589/jpts.24.1133. [41] H.-K. Kang, Y. Kim, Y. Chung, and S. Hwang, “Effects of treadmill training with optic flow on balance and gait in individuals following stroke: randomized controlled trials,” Clin. Rehabil. , vol. 26, no. 3, pp. 246–255, Mar. 2012, doi: 10.1177/0269215511419383. [42] A. Mirelman, P. Bonato, and J. E. Deutsch, “Effects of Training With a Robot-Virtual Reality System Compared With a Robot Alone on the Gait of Individuals After Stroke,” Stroke , vol. 40, no. 1, pp. 169–174, Jan. 2009, doi: 10.1161/STROKEAHA.108.516328. [43] G. Morone et al. , “The Efficacy of Balance Training with Video Game-Based Therapy in Subacute Stroke Patients: A Randomized Controlled Trial,” Biomed Res. Int. , vol. 2014, pp. 1–6, 2014, doi: 10.1155/2014/580861. [44] Y.-R. Yang, M.-P. Tsai, T.-Y. Chuang, W.-H. Sung, and R.-Y. Wang, “Virtual reality-based training improves community ambulation in individuals with stroke: A randomized controlled trial,” Gait 106 Table 0.3 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), ROM of both left and right knee joint angle on the sagittal plane (degrees), and number of times each participant kicked the bathtub rim, in both trials, while entering to the bathtub (EB), in Take a Shower virtual challenge “Take a Shower” Virtual Challenge Task Trial Participant Primary Outcomes Secondary Outcomes COM displacement (cm) COM velocity (cm/s) ROM of knee joint angle (sagittal plane) (degrees) Virtual challenge Score AP ML AP ML Left Right Number of times the user kicked the bathtub rim Min Max Min Max EB T1 P1 101.37 35.13 - 24.71 52.64 - 39.76 35.52 110.17 108.48 21 P2 84.01 18.07 - 10.97 55.70 - 12.58 10.15 109.56 105.69 3 P3 85.14 16.57 - 15.29 55.49 - 29.90 24.86 106.85 112.83 5 P4 89.23 15.61 - 6.50 50.78 - 16.19 16.04 112.77 113.84 1 P5 92.36 26.42 - 12.65 83.71 - 24.97 29.24 109.63 90.91 0 P6 81.20 20.48 - 7.86 31.84 -12.39 13.83 105.26 113.01 1 T2 P1 91.70 22.83 - 16.94 43.74 - 21.82 24.90 108.92 106.91 1 P2 79.24 15.54 - 14.05 56.42 - 10.29 18.78 114.99 108.87 1 P3 78.99 17.70 - 9.92 70.98 - 18.59 26.11 100.10 114.15 1 P4 94.68 29.59 -14.06 62.21 - 30.21 32.18 114.69 114.79 1 P5 93.73 14.13 - 8.13 76.15 - 25.35 21.65 107.54 96.21 0 P6 90.23 21.03 - 7.30 50.83 - 16.38 11.36 103.66 115.68 1 107 Table 0.4 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), ROM of both left and right knee joint angle on the sagittal plane (degrees), and the amount of time the sound played (s) 10 s being the total, in both trials, while each participant is inside the bathtub (IB), in Take a Shower virtual challenge “Take a Shower” Virtual Challenge Task Trial Participant Primary Outcomes Secondary Outcomes COM displacement (cm) COM velocity (cm/s) ROM of knee joint angle (sagittal plane) (degrees) Virtual Challenge Score AP ML AP ML Left Right Sound playing duration inside the bathtub Min Max Min Max IB T1 P1 9.51 14.63 - 7.42 7.21 - 13.00 20.72 19.48 93.47 10.00 P2 9.77 17.78 - 10.10 9.97 - 15.85 21.63 92.81 27.42 10.00 P3 5.50 10.09 - 7.18 7.89 - 12.81 11.62 17.93 103.33 2.58 P4 9.44 15.75 - 5.71 10.56 - 10.42 20.13 28.12 91.90 4.60 P5 7.34 10.50 - 8.93 7.23 - 6.76 12.13 10.95 80.38 10.00 P6 7.72 16.16 - 14.72 12.12 - 6.16 15.04 32.25 87.49 3.91 T2 P1 8.61 14.52 - 8.91 11.10 - 11.79 18.35 23.95 95.67 4.53 P2 9.94 16.58 - 6.53 9.63 - 10.37 19.52 97.22 31.33 10.00 P3 6.75 8.42 - 7.72 5.63 - 6.64 10.20 28.16 96.73 10.00 P4 13.88 24.00 - 8.21 9.09 - 34.66 41.48 44.15 79.74 1.74 P5 7.76 11.37 - 5.13 8.46 - 7.79 13.09 12.02 73.14 10.00 P6 5.65 10.94 - 4.18 6.53 - 8.52 8.48 11.15 67.21 10.00 108 Table 0.5 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), and the number of ingredients caught, in both trials, by each participant, in Cooking virtual challenge “Cooking” Virtual Challenge Trial Participant Primary Outcomes Secondary Outcome COM displacement (cm) COM velocity (cm/s) Virtual Challenge Score AP ML AP ML Number of ingredients caught in kitchen’s shelves Min Max Min Max T1 P1 18.84 10.50 - 10.04 6.86 - 6.22 7.22 6 P2 7.54 9.77 - 10.65 7.79 - 9.67 9.59 6 P3 7.24 5.90 - 5.93 7.21 - 8.46 7.02 6 P4 9.20 25.96 - 9.41 12.27 - 17.22 18.04 6 P5 15.15 17.14 - 9.28 15.17 - 14.70 16.24 6 P6 16.67 14.65 - 7.11 16.75 - 11.50 10.16 6 T2 P1 17.23 6.45 - 7.98 5.63 - 7.03 7.12 6 P2 8.39 8.65 - 6.69 7.45 - 5.92 7.61 6 P3 8.34 8.50 - 7.01 5.95 - 10.98 11.64 6 P4 12.25 26.79 - 9.62 11.01 - 18.20 19.52 6 P5 19.51 22.21 - 8.90 8.19 - 13.88 20.11 6 P6 18.23 14.70 - 9.39 9.68 - 10.46 10.59 6 109 Table 0.6 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), and ROM of both left and right knee joint angle on the sagittal plane (degrees) of each participant while sitting, in both trials, in Watch Tv virtual challenge “Watch TV” Virtual Challenge Task Trial Participant Primary Outcomes Secondary Outcomes COM displacement (cm) COM velocity (cm/s) ROM of knee joint angle (sagittal plane) (degrees) AP ML AP ML Left Right Min Max Min Max Stand-to-sit T1 P1 45.54 3.96 - 27.25 1.38 - 4.52 5.37 67.53 67.93 P2 40.67 4.16 - 42.85 1.65 - 4.00 2.25 78.32 81.95 P3 25.06 2.23 - 16.23 0.94 - 7.12 4.29 84.59 83.58 P4 31.52 3.51 - 19.80 1.83 - 7.93 6.43 99.30 97.14 P5 36.89 1.65 - 24.70 3.09 - 3.89 5.10 65.61 64.22 P6 30.80 5.35 - 18.95 1.87 - 5.78 2.56 70.85 83.16 T2 P1 44.59 6.65 - 24.17 3.67 - 12.99 7.14 75.74 74.80 P2 33.88 3.51 - 23.43 1.69 - 3.70 1.24 81.42 84.15 P3 32.38 5.83 - 28.20 0.74 - 7.02 8.79 80.29 79.52 P4 32.99 3.77 - 21.71 1.96 - 7.54 8.87 100.53 101.70 P5 41.54 3.57 - 26.62 1.69 - 2.61 6.54 63.33 58.27 P6 29.55 3.86 - 19.72 0.53 - 4.67 1.44 73.42 83.35 110 Table 0.7 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), and ROM of both left and right knee joint angle on the sagittal plane (degrees) of each participant while standing, in both trials, in Watch Tv virtual challenge “Watch TV” Virtual Challenge Task Trial Participant Primary Outcomes Secondary Outcomes COM displacement (cm) COM velocity (cm/s) ROM of knee joint angle (sagittal plane) (degrees) AP ML AP ML Left Right Min Max Min Max Sit-to-stand T1 P1 45.73 4.64 - 0.57 36.97 - 8.48 0.73 59.22 57.33 P2 37.88 4.50 0.19 33.64 - 6.77 1.99 67.28 75.27 P3 23.90 2.44 0.48 28.67 - 5.41 4.35 82.04 80.00 P4 28.74 3.03 1.20 31.05 - 2.60 6.13 89.22 88.86 P5 33.76 1.91 3.09 33.42 - 3.99 6.64 64.69 62.76 P6 32.35 3.72 0.99 30.21 - 6.98 3.32 67.50 80.88 T2 P1 43.12 3.99 - 0.66 38.97 - 5.78 2.34 64.83 66.42 P2 35.53 2.25 - 3.35 46.44 - 4.37 3.78 65.67 75.74 P3 30.20 1.71 - 0.76 34.67 - 5.16 4.86 67.19 68.80 P4 26.86 3.70 1.80 30.68 - 3.46 9.45 89.94 90.53 P5 39.13 1.90 1.25 47.23 - 4.69 5.59 60.95 54.55 P6 28.86 3.59 - 2.72 28.95 - 1.15 3.58 72.72 81.32 111 Table 0.8 – Mean and standard deviation of COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), ROM of both left and right knee joint angle on the sagittal plane (degrees), stride time (s), and stride length (m) of each participant while walking, in both trials, and if he turn on the television, in Watch Tv virtual challenge “Watch TV” Virtual Challenge Task Trial Participant Primary Outcomes Secondary Outcomes COM displacement ( ± σ) (cm) COM velocity ( ± σ) (cm/s) ROM of knee joint angle (sagittal plane) ( ± σ) (degrees) Temporal and spatial gait parameters ( ± σ) Virtual challenge Score AP ML AP ML Left Right Stride Time (s) Stride Length (cm) Television state achieved Min Max Min Max Walking T1 P1 70.33 ± 3.64 6.95 ± 1.26 31.02 ± 5.06 54.93 ± 3.78 - 22.21 ± 3.06 11.87 ± 3.12 54.30 ± 2.65 53.85 ± 2.75 1.74 ± 0.12 70.98 ± 4.56 On P2 67.32 ± 6.69 7.88 ± 1.95 38.96 ± 6.21 57.09 ± 4.45 - 20.22 ± 4.92 19.78 ± 3.67 47.87 ± 1.91 47.50 ± 3.85 1.42 ± 0.05 67.85 ± 7.81 On P3 83.87 ± 2.47 12.61 ± 2.63 46.99 ± 2.82 70.44 ± 3.76 - 18.90 ± 2.75 27.82 ± 4.51 58.64 ± 1.35 58.12 ± 0.32 1.48 ± 0.08 84.94 ± 1.33 On P4 66.94 ± 8.11 8.30 ± 2.37 43.76 ± 6.53 67.41 ± 11.49 - 22.93 ± 1.92 16.89 ± 8.30 49.82 ± 4.81 52.91 ± 3.15 1.27 ± 0.10 66.74 ± 6.98 On P5 84.75 ± 9.79 11.66 ± 4.63 54.69 ± 9.66 79.56 ± 8.00 - 12.19 ± 5.64 26.36 ± 5.79 53.90 ± 1.78 51.26 ± 3.12 1.29 ± 0.05 84.28 ± 9.15 On P6 69.04 ± 35.44 7.28 ± 2.60 40.46 ± 5.18 59.64 ± 8.71 - 25.58 ± 2.10 7.44 ± 12.34 51.09 ± 14.59 52.38 ± 29.19 1.36 ± 0.68 66.91 ± 37.80 On 112 “Watch TV” Virtual Challenge Task Trial Participant Primary Outcomes Secondary Outcomes COM displacement ( ± σ) (cm) COM velocity ( ± σ) (cm/s) ROM of knee joint angle (sagittal plane) ( ± σ) (degrees) Temporal and spatial gait parameters ( ± σ) Virtual challenge Score AP ML AP ML Left Right Stride Time (s) Stride Length (cm) Did the user turn on the Tv? Min Max Min Max Walking T2 P1 72.27 ± 10.35 7.63 ± 1.89 36.21 ± 3.73 63.19 ± 9.17 - 14.83 ± 3.69 14.05 ± 5.13 55.66 ± 2.38 56.86 ± 2.22 1.60 ± 0.12 80.34 ± 6.45 On P2 54.82 ± 22.80 6.41 ± 1.54 37.47 ± 8.89 53.43 ± 13.53 - 20.88 ± 4.33 11.70 ± 8.10 44.43 ± 8.19 38.88 ± 17.00 1.21 ± 0.43 54.07 ± 26.16 On P3 88.58 ± 2.66 9.94 ± 3.01 51.00 ± 7.45 79.39 ± 4.66 - 18.78 ± 3.90 26.72 ± 4.85 60.25 ± 1.25 58.36 ± 1.45 1.42 ± 0.12 89.88 ± 0.05 On P4 71.38 ± 10.76 6.90 ± 1.04 47.78 ± 6.25 70.54 ± 9.19 - 23.20 ± 3.13 13.86 ± 5.33 53.84 ± 3.71 53.74 ± 2.02 1.24 ± 0.05 70.60 ± 12.08 On P5 89.85 ± 6.79 7.78 ± 2.33 61.75 ± 5.92 85.66 ± 2.55 - 19.44 ± 5.33 22.52 ± 4.69 54.53 ± 1.81 53.45 ± 2.81 1.23 ± 0.03 88.88 ± 6.70 On P6 89.38 ± 6.74 6.24 ± 0.71 46.43 ± 6.46 72.00 ± 2.40 - 23.69 ± 0.37 14.36 ± 1.38 58.34 ± 0.37 68.05 ± 1.59 1.55 ± 0.04 90.07 ± 5.57 On 113 Table 0.9 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), ROM of both left and right knee joint angle on the sagittal plane (degrees), and the BBS score of each participant, for BBS tasks 1,4,12, and 14, respectively, before and after VR training BBS BBS Task Participant Primary Outcomes Secondary Outcomes Clinical Scale Assessment COM displacement (cm) COM velocity (cm/s) ROM of knee joint angle (degrees) BBS Score AP ML AP ML Left Right Min Max Min Max Sit-to-stand Before P1 44.37 3.01 - 3.95 47.85 - 3.15 0.67 62.74 69.06 4 P2 47.40 2.44 - 5.83 60.57 - 6.45 5.06 86.72 86.78 4 P3 40.93 2.77 - 3.03 48.46 - 2.63 5.58 79.83 74.21 4 P4 40.71 2.04 - 6.74 44.02 - 5.16 2.06 95.38 94.34 4 P5 42.95 3.31 - 13.17 48.25 - 3.38 1.29 68.87 58.59 4 P6 37.05 1.87 - 8.78 39.55 - 2.71 2.80 70.94 88.41 4 After P1 39.45 1.42 - 5.71 46.61 - 3.05 2.32 64.86 68.90 4 P2 42.02 3.66 - 4.30 64.61 - 5.70 3.06 86.36 83.69 4 P3 29.34 3.92 - 2.78 34.10 - 1.61 7.68 82.96 85.87 4 P4 42.19 2.46 - 29.56 5.36 - 4.40 3.26 94.47 95.81 4 P5 41.06 2.28 - 9.70 47.59 - 4.65 4.77 64.72 59.18 4 P6 31.23 1.56 - 7.96 35.00 - 1.73 3.09 80.68 89.16 4 Stand-to-sit Before P1 42.92 0.54 - 38.65 3.68 - 7.93 1.60 61.41 69.42 4 P2 44.46 4.63 - 56.28 2.33 - 4.78 1.93 80.41 81.38 4 P3 36.53 7.15 - 43.76 0.99 - 11.58 2.50 79.01 76.08 4 P4 42.66 3.04 - 37.03 1.38 - 2.48 1.63 95.26 94.54 4 P5 41.08 3.48 - 35.67 3.50 - 3.48 2.20 68.02 60.54 4 P6 32.70 4.86 - 25.74 2.95 - 1.29 3.98 70.04 86.69 4 After P1 40.98 1.22 - 49.18 4.05 - 2.21 2.34 62.96 65.11 4 P2 39.78 1.64 - 49.38 3.70 - 2.70 3.86 83.53 80.65 4 P3 29.49 5.43 -36.63 1.42 - 9.42 2.16 84.66 87.31 4 P4 35.01 2.75 - 24.15 3.57 - 5.12 1.94 96.59 94.68 4 P5 39.29 2.63 - 42.02 6.19 - 3.83 2.68 65.59 63.89 4 P6 30.31 4.73 - 27.99 3.05 - 3.84 1.50 79.57 89.33 4 114 BBS BBS Task Participant Primary Outcomes Secondary Outcomes Clinical Scale Assessment COM displacement (cm) COM velocity (cm/s) ROM of knee joint angle (degrees) BBS Score AP ML AP ML Left Right Min Max Min Max Place alternate foot on step Before P1 6.10 18.88 - 7.32 9.52 - 27.81 25.65 67.32 70.33 4 P2 6.10 14.42 - 8.23 10.48 - 18.58 18.25 67.58 59.61 4 P3 5.23 16.52 - 6.66 5.76 - 18.87 19.27 83.30 77.83 4 P4 7.10 18.23 - 9.11 11.19 - 18.93 20.90 77.09 76.02 4 P5 9.09 14.79 - 8.21 11.24 - 19.67 20.12 67.88 60.30 4 P6 7.00 15.03 - 8.69 10.03 - 14.79 16.61 68.37 76.38 4 After P1 6.74 15.05 - 6.39 7.14 - 21.33 19.53 68.72 73.23 4 P2 5.44 15.70 - 7.87 9.45 - 20.61 22.77 65.29 62.09 4 P3 10.95 15.99 - 6.69 6.43 - 20.44 21.45 93.70 92.96 4 P4 8.68 15.46 - 10.34 12.38 - 18.29 18.81 74.14 78.41 4 P5 6.15 15.72 - 5.84 8.93 - 17.03 19.37 72.32 64.58 4 P6 6.45 15.62 - 7.23 10.78 - 16.82 17.53 77.89 83.07 4 Standing on one leg Before P1 6.02 13.46 - 2.80 4.93 - 18.38 1.82 89.94 7.38 4 P2 6.82 13.38 - 6.16 6.60 - 15.20 16.37 96.06 19.18 4 P3 2.76 9.40 - 1.56 2.70 - 11.21 1.58 92.42 4.83 4 P4 2.61 2.30 - 4.59 13.95 - 6.34 14.16 3.84 49.73 4 P5 3.83 9.91 - 4.78 3.34 - 3.44 11.89 3.77 69.53 4 P6 5.01 7.75 - 3.23 6.20 - 1.27 11.30 5.79 80.50 4 After P1 8.79 12.37 - 2.06 9.02 - 13.96 6.73 91.58 8.15 4 P2 4.24 11.04 - 3.77 9.57 - 17.50 2.90 83.49 9.91 4 P3 4.77 7.16 - 0.98 4.06 - 11.80 1.75 104.84 2.99 4 P4 3.66 6.87 - 3.04 6.15 - 3.28 8.62 17.58 86.75 4 P5 3.37 10.81 - 4.10 8.17 - 2.11 15.06 6.48 73.06 4 P6 4.94 6.25 - 3.17 5.06 - 2.15 9.44 8.55 79.13 4 115 Table 0.10 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), maximum hand forward position (cm), and the BBS score of each participant, for BBS task 8, before and after VR training BBS BBS Task Participant Primary Outcomes Secondary Outcomes Clinical Scale Assessment COM displacement (cm) COM velocity (cm/s) Maximum hand position (cm) BBS Score AP ML AP ML Forward Min Max Min Max Reaching forward Before P1 11.91 1.20 - 12.52 11.04 - 1.32 1.36 62.69 4 P2 8.14 5.07 - 9.74 8.65 - 2.63 5.14 72.40 4 P3 10.63 3.15 - 14.25 8.56 - 3.84 4.17 59.57 4 P4 5.99 4.04 - 7.92 3.81 - 2.97 5.41 68.19 4 P5 10.04 2.64 - 12.86 5.05 - 2.92 3.72 67.75 4 P6 11.93 3.10 - 12.58 5.35 - 2.94 3.32 75.29 4 After P1 10.26 4.83 - 15.27 10.74 - 3.30 7.48 64.52 4 P2 6.27 3.87 - 6.52 7.82 - 3.37 3.67 68.73 4 P3 8.97 1.40 - 8.58 9.95 - 6.35 1.18 57.56 4 P4 5.21 2.58 - 6.74 4.19 - 2.87 3.79 66.44 4 P5 8.31 3.20 - 6.38 6.22 - 2.20 2.23 69.22 4 P6 11.17 3.19 - 9.73 9.22 - 1.93 4.11 75.87 4 122 Table 0.13 - Mean and standard deviation of COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), ROM of both left and right knee joint angle on the sagittal plane (degrees), stride time (s), and stride length (m) of each participant while performing the TUG walking task, before and after VR training TUG Task Trial Participant Primary Outcomes Secondary Outcomes COM displacement ( ± σ) (cm) COM velocity ( ± σ) (cm/s) ROM of knee joint angle (sagittal plane) ( ± σ) (degrees) Temporal and spatial gait parameters ( ± σ) AP ML AP ML Left Right Stride Time (s) Stride Length (cm) Min Max Min Max Walking Before P1 118.52 ± 8.17 16.20 ± 6.82 91.69 ± 4.12 120.27 ± 3.54 - 14.75 ± 23.75 17.68 ± 20.92 61.83 ± 0.92 69.32 ± 0.67 1.16 ± 0.04 118.61 ± 3.40 P2 106.03 ± 13.99 15.48 ± 7.02 78.86 ± 23.31 120.13 ± 12.25 - 24.42 ± 31.19 17.54 ± 18.66 62.45 ± 1.14 57.97 ± 4.00 1.09 ± 0.03 103.62 ± 24.11 P3 114.51 ± 7.54 25.89 ± 16.75 73.99 ± 5.15 107.28 ± 3.18 - 3.27 ± 11.96 38.51 ± 19.25 66.09 ± 2.53 60.56 ± 1.64 1.32 ± 0.04 114.48 ± 6.84 P4 123.77 ± 2.29 13.22 ± 6.30 89.37 ± 5.32 121.90 ± 4.96 - 18.75 ± 18.30 5.50 ± 17.58 62.55 ± 3.22 62.93 ± 1.11 1.21 ± 0.04 125.02 ± 3.84 P5 118.44 ± 17.36 10.70 ± 9.58 96.77 ± 9.93 131.69 ± 5.43 - 23.87 ± 5.29 14.22 ± 16.17 58.93 ± 2.81 53.36 ± 5.56 1.05 ± 0.09 115.89 ± 24.26 P6 110.32 ± 15.31 10.49 ± 5.55 84.75 ± 14.39 119.91 ± 4.96 - 25.66 ± 14.91 13.20 ± 12.91 61.75 ± 2.17 72.22 ± 2.07 1.11 ± 0.04 111.32 ± 18.37 123 TUG Task Trial Participant Primary Outcomes Secondary Outcomes COM displacement ( ± σ) (cm) COM velocity ( ± σ) (cm/s) ROM of knee joint angle (sagittal plane) ( ± σ) (degrees) Temporal and spatial gait parameters ( ± σ) AP ML AP ML Left Right Stride Time (s) Stride Length (cm) Min Max Min Max Walking After P1 118.63 ± 5.36 9.27 ± 4.47 90.43 ± 1.15 123.42 ± 3.63 - 20.19 ± 14.86 7.93 ± 7.07 65.26 ± 1.09 68.10 ± 0.88 1.15 ± 0.03 121.95 ± 3.31 P2 106.61 ± 8.17 21.34 ± 13.63 81.32 ± 10.93 106.05 ± 6.61 - 39.24 ± 26.78 4.03 ± 19.08 58.12 ± 1.83 58.71 ± 0.79 1.16 ± 0.02 109.07 ± 6.48 P3 112.30 ± 1.86 19.31 ± 8.32 70.76 ± 2.67 103.13 ± 1.03 - 9.83 ± 7.42 26.56 ± 21.81 61.73 ± 1.62 62.53 ± 2.75 1.37 ± 0.02 112.84 ± 3.94 P4 116.27 ± 14.85 20.19 ± 14.70 85.96 ± 12.34 115.50 ± 18.57 - 32.70 ± 16.69 - 4.51 ± 21.43 59.67 ± 2.50 59.86 ± 1.08 1.19 ± 0.06 118.46 ± 17.50 P5 118.96 ± 7.01 11.31 ± 6.36 97.20 ± 8.74 126.01 ± 6.49 - 19.88 ± 9.28 19.21 ± 11.45 59.67 ± 1.98 56.17 ± 3.47 1.09 ± 0.02 120.92 ± 10.37 P6 110.03 ± 5.02 10.20 ± 5.86 79.73 ± 4.84 108.32 ± 0.78 - 15.55 ± 10.96 16.41 ± 13.15 63.08 ± 2.91 72.21 ± 0.91 1.19 ± 0.01 112.12 ± 7.12 124 Table 0.14 - COM displacement on AP and ML directions (cm), maximum and minimum COM velocity on AP and ML directions (cm/s), and ROM of both left and right knee joint angle on the sagittal plane (degrees) of each participant while performing the TUG sitting task, before and after VR training TUG Task Trial Participant Primary Outcomes Secondary Outcomes COM displacement (cm) COM velocity (cm/s) ROM of knee joint angle (sagittal plane) (degrees) AP ML AP ML Left Right Min Max Min Max Stand-to-sit Before P1 29.86 1.12 - 32.77 1.95 - 3.87 2.71 41.52 43.92 P2 39.29 5.65 - 40.79 1.72 - 3.07 16.15 80.54 83.87 P3 26.17 8.90 - 34.34 0.87 - 9.86 0.73 65.92 66.27 P4 35.81 7.16 - 33.33 2.58 - 27.68 8.69 72.05 71.29 P5 41.31 3.04 - 45.06 2.43 - 12.19 2.69 47.42 37.48 P6 35.58 4.70 - 25.68 1.33 - 10.15 2.15 66.19 69.82 After P1 33.10 3.75 - 30.22 3.03 - 8.18 5.02 59.94 53.09 P2 42.86 5.91 - 44.55 0.82 - 18.21 4.80 80.11 77.48 P3 27.86 6.77 - 34.94 0.87 - 11.17 1.10 64.50 66.55 P4 32.81 3.19 - 34.52 1.27 - 3.44 13.76 62.59 55.39 P5 47.92 6.95 - 46.73 2.17 - 15.30 6.22 64.61 61.32 P6 33.70 9.21 - 29.64 1.57 - 15.83 1.31 76.20 83.19 125 APPENDIX II – NORMALITY AND HOMOGENEITY STATISTICAL TESTS Tables 0.15 to 0.18, 0.19 and 0.20, and 0.21 present the p-values from the normality tests (Kolmogorov-Smirnov) for the primary outcome measures: COM displacement (Disp), maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for the four virtual challenges, the sensor-based assessment of the two clinical tests (BBS and TUG), and for the BSB and TUG clinical assessment scores, respectively. In the same order, for the metrics following a normal distribution, Tables 0.22 to 0.25, 0.26 and 0.27, and 0.28, present their p-values after the homogeneity test (Levene). Table 0.15 - The p-values of the normality tests for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for CS, and CF motor tasks of the virtual challenge “Fruit Catcher”, considering a significance level of 5 %. The most representative results ( p -value < 0.05) appear in bold Tests of Normality Catch Fruit Kolmogorov-Smirnov Sig. Motor tasks T1 T2 Climbing Stairs DispAP_CS .200* .200* DispML_CS .200* .200* VelMinAP_CS .200* .200* VelMaxAP_CS .200* .200* VelMinML_CS .200* . 143 VelMaxML_CS .200* .200* Catching Fruits DispAP_CF .200* .200* DispML_CF .200* .200* VelMinAP_CF .200* .200* VelMaxAP_CF .119 .200* VelMinML_CF .200* .200* VelMaxML_CF .200* .150 *. This is a lower bound of the true significance. a. Lilliefors Significance Correction 126 Table 0.16 - The p-values of the normality tests for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for EB, and standing on one leg inside the bathtub (IB) motor tasks of the virtual challenge “Take a Shower”, considering a significance level of 5 %. The most representative results ( p -value < 0.05) appear in bold Tests of Normality Take a Shower Kolmogorov-Smirnov Sig. Motor tasks T1 T2 Entering in the bathtub DispAP_EB .200* .141 DispML_EB .200* .200* VelMinAP_EB .200* .200* VelMaxAP_EB .059 .200* VelMinML_EB .200* .200* VelMaxML_EB .200* .200* Standing on one leg DispAP_IB .200* .200* DispML_IB .200* .200* VelMinAP_IB .200* .200* VelMaxAP_IB .200* .200* VelMinML_IB .200* .005 VelMaxML_IB .200* .097 *. This is a lower bound of the true significance. a. Lilliefors Significance Correction 127 Table 0.17 - The p-values of the normality tests for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, collecting food from the kitchen shelves of the virtual challenge “Cooking”, considering a significance level of 5 %. The most representative results ( p -value < 0.05) appear in bold Tests of Normality Cooking Kolmogorov-Smirnov Sig. Motor tasks T1 T2 Pick up the food DispAP_Cook .200* .200* DispML_Cook .200* .200* VelMinAP_Cook .142 .200* VelMaxAP_Cook .191 .200* VelMinML_Cook .200* .200* VelMaxML_Cook .197 .200* *. This is a lower bound of the true significance. a. Lilliefors Significance Correction 128 Table 0.18 - The p-values of the normality tests for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for stand-to-sit, sit-to-stand, and walking motor tasks of the virtual challenge “Watch Tv”, considering a significance level of 5 %. The most representative results ( p -value < 0.05) appear in bold Tests of Normality Watch TV Kolmogorov-Smirnov Sig. Motor tasks T1 T2 Stand-to-sit DispAP_Sit .200* .109 DispML_Sit .200* .016 VelMinAP_Sit .200* .200* VelMaxAP_Sit .122 .200* VelMinML_Sit .200* .200* VelMaxML_Sit .200* .200* Sit-to-stand DispAP_Stand .200* .200* DispML_Stand .200* .200* VelMinAP_Stand .200* .200* VelMaxAP_Stand .200* .200* VelMinML_Stand .200* .200* VelMaxML_Stand .200* .200* Walking DispAP_Walk .047 .157 DispML_Walk .101 .200* VelMinAP_Walk .200* .200* VelMaxAP_Walk .200* .200* VelMinML_Walk .200* .200* VelMaxML_Walk .200* .021 *. This is a lower bound of the true significance. a.Lilliefors Significance Correction 129 Table 0.19 - The p-values of the normality tests for COM displacement (Disp), and maximum (Max) and minimum (Min) COM velocity (Vel), on AP and ML directions, for BBS motor tasks, considering a significance level of 5 %. The most representative results ( p -value < 0.05) appear in bold Tests of Normality BBS Kolmogorov-Smirnov Sig. Motor tasks Pre Post Sit-to-stand DispAP_BBS1 .200* .108 DispML_BBS1 .200* .200* VelMinAP_BBS1 .200* .024 VelMaxAP_BBS1 .066 .200* VelMinML_BBS1 .094 .200* VelMaxML_BBS1 .200* .055 Standing unsupported DispAP_BBS2 .200* .009 DispML_BBS2 .200* .082 VelMinAP_BBS2 .028 .200* VelMaxAP_BBS2 .191 .188 VelMinML_BBS2 .000 .014 VelMaxML_BBS2 .092 .200* Sitting unsupported DispAP_BBS3 .025 .200* DispML_BBS3 .200* .074 VelMinAP_BBS3 .200* .162 VelMaxAP_BBS3 .072 .188 VelMinML_BBS3 .200* .200* VelMaxML_BBS3 .001 .200* *. This is a lower bound of the true significance. a. Lilliefors Significance Correction 130 Tests of Normality BBS Kolmogorov-Smirnov Sig. Motor tasks Pre Post Stand-to-sit DispAP_BBS4 .200* .200* DispML_BBS4 .200* .200* VelMinAP_BBS4 .200* .200* VelMaxAP_BBS4 .200* .200* VelMinML_BBS4 .200* .198 VelMaxML_BBS4 .200* .200* Transfers DispAP_BBS5 .200* .101 DispML_BBS5 .200* .200* VelMinAP_BBS5 .200* .200* VelMaxAP_BBS5 .200* .200* VelMinML_BBS5 .200* .200* VelMaxML_BBS5 .130 .043 Standing with eyes closed DispAP_BBS6 .200* .200* DispML_BBS6 .052 .200* VelMinAP_BBS6 .200* .200* VelMaxAP_BBS6 .143 .200* VelMinML_BBS6 .043 .200* VelMaxML_BBS6 .200* .200* *. This is a lower bound of the true significance. a. Lilliefors Significance Correction 131 Tests of Normality BBS Kolmogorov-Smirnov Sig. Motor tasks Pre Post Standing with feet together DispAP_BBS7 .200* .200* DispML_BBS7 .052 .014 VelMinAP_BBS7 .187 .200* VelMaxAP_BBS7 .200* .200* VelMinML_BBS7 .200* .200* VelMaxML_BBS7 .200* .004 Reaching forward DispAP_BBS8 .200* .200* DispML_BBS8 .200* .200* VelMinAP_BBS8 .069 .200* VelMaxAP_BBS8 .200* .200* VelMinML_BBS8 .200* .047 VelMaxML_BBS8 .200* .200* Pick up an object DispAP_BBS9 .200* .200* DispML_BBS9 .058 .188 VelMinAP_BBS9 .200* .200* VelMaxAP_BBS9 .200* .063 VelMinML_BBS9 .200* .072 VelMaxML_BBS9 .200* .200* *. This is a lower bound of the true significance. a. Lilliefors Significance Correction