Abstract
This perspective article explores the current challenges and future potential of artificial intelligence (AI)-assisted immersive virtual reality (VR) for clinical applications in movement-based physical rehabilitation. It highlights the limitations of current immersive VR and introduces AI as a potential solution to improve the overall VR experience, particularly in the context of physical rehabilitation. AI’s role in content creation, personalizing clinical interventions, achieving the right balance between challenge and flow (where tasks are difficult enough to stay engaging but not so hard that they cause frustration), and optimizing sensory experiences in VR environments is discussed. In addition, the article highlights how AI-driven brain-computer interfaces, motion synthesis techniques, and personalized therapeutic immersive experiences can make VR more engaging and effective in enhancing physical rehabilitation outcomes. Finally, we examine the challenges associated with integrating AI and VR systems in physical rehabilitation, including financial, ethical, and user-related barriers that must be addressed to ensure effective and responsible implementation.
1 Introduction
Over the last decade, immersive virtual reality (VR) technology, characterized by real-time simulation enabling user interaction through multiple sensory channels, has rapidly evolved (). Although VR is now used predominantly for entertainment, it has also come to occupy an increasingly important role in healthcare and the health sciences, where it contributes to a range of therapeutic applications and has demonstrated clinical benefits (; ; Wang et al., 2023). More recently, VR has gained growing attention in the field of physical rehabilitation, offering engaging, interactive environments that can motivate patients, enhance adherence, and support repetitive, task-specific training in a more engaging way — key principles in neurorehabilitation and musculoskeletal recovery (; ; Ye et al., 2023; Rutkowski et al., 2020; ; ; ; Plavoukou et al., 2025). Evidence from meta-analysis, systematic and scoping reviews suggests that VR-based rehabilitation can improve motor outcomes and patient engagement (; ; Ye et al., 2023; Rutkowski et al., 2020; ; ; ; Plavoukou et al., 2025).
The success of such therapeutic applications can be influenced by the level of user immersion, which is defined by how closely a system’s input devices (monitoring body responses) and output devices (stimulating sensory perception) align with real-world experiences. Higher fidelity between these components enhance the immersive quality (Slater et al., 2009).
For increased levels of immersion, various input technologies—such as controllers, motion trackers, and brain-controlled systems—capture the user's actions and feed them into the system (). Additionally, output technologies, like headphones, head-mounted displays, and vibration-based devices offer users extrinsic feedback to engage user senses (). In an fully immersive experience, complete attention is required, involving all the senses and allowing users to interact within the virtual world just as they would in the real world (). The degree of immersion is impacted by the sense of presence (psychological perception of being present in the virtual environment), the level of realism (alignment between the user’s expectations and the actual experience in the virtual world), and plausibility/authenticity of the experience (the degree at which the user perceives immersion as authentic) (Slater et al., 2009; ; ; Suh and Prophet, 2018). In our view, artificial intelligence (AI) represents a promising avenue to further enhance these elements by enabling systems to dynamically adapt to users’ experience.
The integration of AI in medicine has been a research topic since the 1950s (; ), with its first surgical application in diagnosing acute abdominal pain in 1976 (). Since then, AI has significantly evolved, expanding into subtypes ranging from narrow AI, focused on specific tasks, to the ambitious pursuit of strong AI, which aims to match or surpass human intelligence (). Its uses now span a wide range of clinical applications, from diagnosis and prognosis to assisting in decision-making (Singareddy et al., 2023).
AI’s ability to process vast amounts of real-time data (particularly in analyzing motor performance, psychophysiological responses, and behavioral patterns), may enable the creation of adaptive VR environments capable of adjusting therapeutic tasks to individual needs (; ; ). Unlike conventional adaptive systems, which often rely on rule-based approaches, AI-based approaches are designed to learn from large and complex datasets, identify patterns, and integrate multiple sources of patient data to support more individualized rehabilitation strategies (). These capabilities have the potential to support a more flexible, scalable, and individualized approach to rehabilitation than conventional rule-based adaptation alone.
The integration of AI into immersive VR systems therefore represents a promising step toward more adaptive and personalized rehabilitation. In this perspective, we argue that the most transformative role of AI in VR-based rehabilitation lies in its capacity to analyze behavioral, motor, and physiological data in real time to dynamically tailor the intervention based on individuals’ clinical characteristics. However, despite this potential, important translational challenges remain, including system usability, clinical validation, and the integration of AI-driven personalization into routine rehabilitation practice. This article discusses the emerging role of AI-enhanced VR in physical rehabilitation and outlines key research priorities needed to translate these technologies into effective clinical tools. In this perspective, we focus on physical rehabilitation interventions that involve active motor training and movement-based therapy. This article is intended for clinicians, rehabilitation researchers, technology developers, and healthcare decision-makers interested in the future integration of AI-driven immersive VR systems into rehabilitation practice.
2 Reality check: limitations of VR in rehabilitation
Despite its potential, immersive VR faces notable limitations, especially in physical rehabilitation contexts. Although VR experiences delivered through head-mounted display (HMD) systems have been shown in several studies and systematic reviews to improve motor recovery and patient engagement (; ; Ye et al., 2023; Rutkowski et al., 2020; ), several factors still impede their widespread clinical adoption. These obstacles largely stem from economic constraints, limited content availability, and user experience issues that strongly influence adherence and therapeutic effectiveness, as reported in studies examining barriers and facilitators to the adoption of immersive VR in healthcare settings (Sadowski et al., 2026; ; ).
2.1 Economic and content limitations in research settings
Personalizing VR environments can modulate individuals’ emotional experiences by adjusting parameters such as visual and auditory stimulus, interactions, and narratives (). Furthermore, the use of embodiment (sensations of being inside, having, and controlling a body ()) and avatar representation (graphical representation of a user (Peters et al., 1997)) have been shown to influence behavior and potentiate motor rehabilitation (Ventura et al., 2023; ).
However, developing personalized content is resource-intensive and time-consuming. Consequently, many systems rely on broadly designed experiences that often overlook individual preferences, needs, and cultural contexts. This limitation contributes to the wide variability seen in VR applications: while some commercial products offer highly polished visuals, research environments typically focus on experimental content and rely on simpler assets due to limited resources. In this context, visual design can still meaningfully enhance user engagement and enjoyment in academic VR experiences (), especially when combined with rich interactivity and sensory feedback.
Challenges like unrealistic animations and the uncanny valley effect – when virtual characters’ movements or facial expressions feel unnatural – can also impact the realism of the virtual experience (Schindler et al., 2017; ). Additionally, VR applications face usability challenges, especially in niche fields like rehabilitation, where testing and refinement are often limited by costs and time. Addressing these challenges requires a structured approach to VR content development that carefully considers both functionality and cost.
2.2 User experience
Beyond cost and content limitations, user experience issues limit VR’s rehabilitation potential. Cybersickness, resulting from mismatches between visual input and physical movement cues, can lead to dizziness, nausea, and general discomfort (). Cognitive overload, from overwhelming sensory input, may result in confusion and stress if not carefully managed (; ; ; Wenk et al., 2023). Technical issues like glitches and unresponsive controls disrupt immersion and cause frustration (). Additionally, poorly calibrated difficulty levels in VR experiences – whether too simple or too challenging – can lead to disengagement and reduced long-term interest ().
3 AI as a solution
While immersive VR has demonstrated therapeutic benefits in motor and cognitive rehabilitation (; ; Ye et al., 2023; Rutkowski et al., 2020; ), its current limitations often restrict its full clinical impact. Emerging research suggests that AI may help address some of these limitations by enabling real-time adaptation, intelligent feedback, and more efficient content creation. AI can also facilitate real-time data analysis that may support long-term monitoring of motor recovery.
3.1 AI can help with content creation
AI has been used in the gaming industry to generate storylines and virtual environments (; Yannakakis and Hallam, 2009). Interestingly, research shows that AI-generated writings are often indistinguishable from those authored solely by humans, further showcasing AI’s potential for seamless integration into various content creation processes (; ).
AI can generate visual content through Generative Adversarial Networks (GANs), which pair a generator that produces new images with a discriminator that judges how realistic those images appear (Sikka et al., 2024). Although current GANs have resolution limitations (Villegas-Ch et al., 2025; Park et al., 2022), combining them with foveated rendering, which uses eye-tracking to focus on high-resolution areas, can improve the visual experience and reduce cybersickness by directing resources to where the user is looking (; Patney et al., 2016).
AI can also generate audio content with AI music generation assistants enabling the rapid creation of new auditory stimulus (). Additionally, by incorporating natural language processing tools, therapeutic interventions using VR programming, such as patient speech rehabilitation, may enhance the experience. AI could also support realistic motion synthesis by generating plausible, natural, and contextually appropriate human movements in virtual character animation ().
3.2 AI can improve the user experience
AI may also enhance user experience in virtual physical rehabilitation by making interactions more adaptive and personalized. Although clinical evidence remains limited, studies and experimental prototypes have explored how AI-driven systems can dynamically adjust virtual environments based on user behavior or physiological signals. The following sections describe how AI could personalize clinical interventions (Section 3.2.1), prevent cognitive overload and motion sickness (Section 3.2.2), and reduce technical difficulties (Section 3.2.3).
3.2.1 Personalizing clinical interventions and balancing challenge
Tailoring VR rehabilitation to individual needs is often crucial for maximizing the effectiveness of therapeutic outcomes. Adaptive systems driven by machine learning have been proposed as a way to personalize VR experiences in real time, using information collected prior to the intervention (e.g., user’s history) or through formative (e.g., user’s biological signals) and summative (e.g., overall skill improvement) evaluation methods (Figure 1) (; ). Recent studies indicate that continuous analysis of real-time user behaviors may allow AI to dynamically optimize VR experiences, making instant adjustments to maximize immersion and user engagement (; ). However, most of these systems remain at the prototype or experimental stage, and their effectiveness in large-scale clinical rehabilitation settings has yet to be fully established.
FIGURE 1
AI could also dynamically adjust the level of difficulty in VR experiences, ensuring not only an optimal balance between challenge and engagement but also an appropriate level of motor demand. In motor learning, it is essential to maintain a difficulty level that is achievable yet sufficiently demanding, and this level must be adapted to each patient as well as to their progression and moment-to-moment condition, maintaining therapeutic relevance and motor challenge. This may be particularly relevant to serious gaming scenarios, which use game-like simulations for purposes beyond entertainment such as rehabilitation ().
3.2.2 Preventing cognitive overload and motion sickness
Preventing cognitive and sensory overload is essential to ensure patient safety and therapeutic effectiveness. Modern VR head-mounted displays equipped with eye-tracking sensors can assess cognitive strain, visual fatigue, and attentional load during rehabilitation exercises (; Patney et al., 2016). When combined with machine learning, these systems may allow early detection of cognitive overload or motion sickness (Yalcin et al., 2024; ), factors that may hinder patient engagement or motor performance. AI could act as a real-time monitoring system, identifying and classifying adverse reactions, then automatically adjusting pace, or visual stimuli to maintain optimal therapeutic conditions ().
Complementing these adaptive AI capabilities, VR has evolved into a truly multisensory medium, extending beyond sight and sound to include touch, temperature, and even smell, illustrating how such multisensory feedback could be integrated into rehabilitation environments. These effects are achieved using technologies such as vibrotactile motors, Peltier elements, skin stretch actuators, and scent cartridges. For instance, OVR Technology's OMARA device (Vermont, USA) uses scentware cartridges to generate thousands of unique scent combinations, while TESLASUIT (London, UK) integrates haptic feedback via electrical muscle and nerve stimulation, along with motion capture for precise tracking (Teslasuit, 2025). By analyzing sensor data in real time, AI may help fine-tune these multisensory cues to enhance engagement while preventing sensory overload ().
3.2.3 Improving user interaction and overcoming technological barriers
AI also has the potential to support patients, clinicians, and caregivers by reducing the technological and usability challenges often encountered in virtual rehabilitation (). Using natural language processing (NLP), intelligent virtual assistants could potentially guide users through setup and exercises, provide real-time feedback, and assist with technical issues. By interpreting spoken or written input in real time, NLP-based chatbots could provide immediate feedback, clarify user doubts, and personalize responses according to each user’s progress, preferences, and level of digital literacy.
4 AI–VR integration: today and tomorrow
Recent systematic and scoping reviews, presented in Table 1, suggest that the integration of AI with VR technologies may enhance rehabilitation approaches by enabling greater personalization, adaptive feedback, and continuous monitoring of patient performance. However, the available evidence remains limited, and many applications have been explored primarily in experimental or early-stage clinical studies. The following sections describe how AI may contribute to VR-based rehabilitation through mind control of a virtual avatar, motion synthesis, and adaptive game design principles such as flow and just-right challenges.
TABLE 1
| Authors, year | Review type | Scope/context | Population | VR settings | AI models | Key takeaways for rehabilitation |
|---|---|---|---|---|---|---|
| A systematic mapping study | Systematic mapping of 53 studies using AI in neurological physical therapy | Mixed (stroke, PD, ALS, etc.) | 2 studies used VR systems for immersive and feedback-driven rehabilitation | The studies used machine learning-based ANN models | AI and VR technologies make rehabilitation more dynamic and engaging, motivating patients to remain consistent in their recovery | |
| Scoping review | AI home-based virtual rehabilitation | Mixed (various conditions) | 13 studies used VR environments incorporating tools such as smartphone, desktop, and web applications | The AI algorithms applied in the studies primarily included fuzzy rule-based approaches, template matching techniques, and deep neural networks | AI-driven VR rehabilitation improved recovery through personalized feedback, adaptive exercises, and remote monitoring | |
| Sumner et al. (2023) | Systematic review of RCTs | AI-supported physical rehabilitation technology tested in the clinical setting | Mixed (various conditions) | 5 studies used VR environments incorporating equipment such as Kinect cameras, 3D displays, head-mounted displays, balance boards, and drum kits | The studies used machine learning–based AI, including neural networks (RNN, MLP, RBFN, etc.), support vector machines (SVM), fuzzy logic systems, and probabilistic or Bayesian models | VR and AI systems enhance rehabilitation by assessing performance, improving motor and balance functions, and motivating patients through personalized and adaptive exercises |
| Systematic review | Comprehensive review of AI-based machine learning applications for physical rehabilitation | Healthy participants | 1 study used a VR exergame platform integrated with AI for motion prediction | The studies used a machine learning-based XGBoost model | The integration of AI and VR exergames provided accurate motion estimation and made rehabilitation more engaging, enabling personalized and real-time feedback | |
| Said et al. (2022) | Systematic review | VR- and robot-therapy BCI rehabilitation using EEG MRCPs | Mixed (various conditions) | 4 VR–MRCP studies; closed-loop control tasks including walking a virtual avatar and a 2D cursor | The studies used machine learning–based AI models, including classifiers such as SVM variants (e.g., MF-SVM), LDA and its extensions (e.g., LPP-LDA) | MRCP-BCI with VR can induce plasticity and enable movement even with absent motor function; VR can boost engagement and be more cost-effective, but evidence is still limited |
| Systematic review | AI algorithms in VEP-based BCIs (P300, SSVEP) for motor rehabilitation | Mixed (various conditions) | 8 studies integrated VR feedback for BCI-based motor and control tasks | The studies used machine learning–based AI models, including classifiers such as LDA (e.g., LDA, Fisher or Bayesian LDA), SVM, and CCA. | VR-based VEP BCIs enhanced sensory feedback and engagement, achieving high accuracy and ITR, though most remained limited to discrete control and lab settings |
Overview of systematic and scoping reviews on AI-driven VR physical rehabilitation technologies.
Abbreviations: Artificial Intelligence (AI); Amyotrophic lateral sclerosis (ALS); Artificial Neural Network (ANN); Brain–Computer Interface (BCI); Canonical Correlation Analysis (CCA); Electroencephalography (EEG); Information Transfer Rate (ITR); Locality Preserving Projection (LPP); Matching Filter (MF); Movement-Related Cortical Potential (MRCP); Multilayer Perceptron (MLP); P300 Event-Related Potential (P300); Parkinson’s disease (PD); Radial Basis Function Network (RBFN); Randomised controlled trials (RCTs); Recurrent Neural Network (RNN); Steady State Visually Evoked potentials (SSVEP); Support Vector Machine (SVM); Three-Dimensional (3D); Two-Dimensional (2D); Visual-Evoked Potential (VEP); Virtual Reality (VR); eXtreme Gradient Boosting (XGBoost).
4.1 Mind control of a virtual avatar
Recent advances have led to the integration of AI in brain–computer interface (BCI) systems, enabling individuals with motor impairments (such as stroke or spinal cord injury patients) to control virtual avatars through neural activity in immersive VR settings (Said et al., 2022; ). By translating brain signals into avatar movements, BCIs combined with AI could allow patients to perform interactive rehabilitation exercises that support motor recovery and strength. When paired with motor imagery (e.g., the mental simulation of movement without physical execution (Rulleau and Toussaint, 2014), this approach may promote gradual re-engagement in movement while reducing fear or pain, as overlapping brain areas are activated during both imagined and actual movements (; ). Systematic reviews and meta-analyses suggest that motor imagery can enhance motor function, strength, and learning (; ; Singer et al., 2024; Yan et al., 2025; ). In this framework, BCIs serve as a bridge between neural activity and the virtual environment, while AI decodes and adapts signals in real time to deliver personalized feedback and potentially improve rehabilitation outcomes (Said et al., 2022; ).
4.2 Motion synthesis
Motion synthesis methods offer an alternative approach to animating avatars and virtual agents in virtual environments, with the aim of reducing reliance on traditional, labor-intensive animation techniques. Recent progress in motion synthesis methods () have demonstrated the potential to generate complex movements in real-time that are perceived by users to be more natural and more akin to the movements of real-life people they interact with.
In the field of physical rehabilitation, the increased realism made possible by motion synthesis has two distinct advantages. First, realistic and natural character motion in the virtual environment has been shown to strengthen social presence (). A pilot study, Shah et al. (2023), found that the implementation of a prosocial element in immersive VR “exergame” rehabilitation programs promoted patient adherence, motivation, and enjoyment when compared to individual VR rehabilitation programs (Shah et al., 2023).
The second advantage of more realistic motion is that users can train in VR using movements and visual cues similar to those in the physical world. Current evidence supports the feasibility and relevance of using such cues to study gaze, locomotion, pedestrian interaction, and movement assessment in controlled or proof-of-concept settings (; ; Winkler et al., 2022; Wei et al., 2019). Such naturalistic motion also reduces the uncanny valley effect, making virtual characters more believable and engaging. Motion synthesized from large datasets of real interactions increases both realism and the real-world transferability of acquired skills. This is especially important for pedestrian and crowd interaction, virtual therapists, and virtual training adversaries, where existing studies provide experimental or methodological support, but clinical effectiveness and functional transfer still need to be confirmed in rehabilitation trials (; ; Winkler et al., 2022; Wei et al., 2019).
Motion synthesis can also be applied to the self-avatar of the user, to animate its body with few markers (Winkler et al., 2022) or, more experimentally, to display movements that are adjusted in real-time to either improve or worsen the quality of the movement that was produced. Such changes in the representation of one’s movements have been shown to be useful in correcting a user’s movements by leading them to either follow their avatar () or correct for perceived deficits (), and are consistent with emerging clinical work on immersive virtual embodiment for pain and motor rehabilitation, including shoulder pain rehabilitation (; ).
4.3 Flow and just-right challenges in serious gaming
Serious games designed for rehabilitation aim to promote engagement and motor learning through structured, goal-oriented play. Serious games can be defined as digital games with the following elements: a software component, educational programming, and goals for the user, which can include problem solving or skill acquisition (). By inducing a strong feeling of presence, VR-based serious games may facilitate the emergence of flow (Riva et al., 2004). Flow, defined as a mental state of deep immersion, focused engagement, and altered time perception (), can influence the perceived effort during physical rehabilitation task (). Its consideration in the design of serious games design may therefore support engagement in therapy, although its direct impact on rehabilitation success depends on the population, task, and intervention context ().
The emergence of the state of flow is facilitated when there is a balance between task difficulty and one’s skill level, maximizing one’s abilities. However, for motor skill learning growth, therapists are aiming to provide interventions at a task difficulty that is just above one’s skill level. This specific level is called the just-right challenge – neither too hard nor too easy – to keep both motivation and efforts high (). Interestingly, Poulsen et al. (2006) described commonalities between flow and just-right challenge. While flow is a perfect balance between skill and difficulty that can be broken by failure, just-right challenge is slightly above one’s abilities, leading to necessary failure that will lead to progress in motor learning. Both concepts share interesting characteristics, such as continuity of effort and motivation to perform a task and improve.
In studies using serious games research, different mechanics have been explored to help maintain users’ engagement and support flow-like experiences (). Schrader et al. (2017) highlight the importance of game characteristics depending on the emotional states of the player. They define this adaptability as a player-centered approach which allows users to adjust the game mechanics and the representation modes to better learning, experience, and performance. The interest of this adaptability based on psychophysiological states is to not have a unique player archetype model, but to have the experience tailored to each. The importance of flow rehabilitation-oriented serious games has also been discussed in several studies (; Petsani et al., 2022; Verschueren et al., 2019).
Yet, the state of flow is still hard to measure as a subjective perception of experience. However, some objective measures have been explored to properly identify these states, ranging from non-verbal behavior to physiological responses that can be collected and analyzed in real time (). AI algorithms could eventually be applied to analyze psychophysiological responses in real-time and detect the conditions needed for the emergence of the state of flow. This may allow for dynamic adaptation of the virtual environment or task to suit the users’ needs.
5 Challenges of AI and VR systems in physical rehabilitation
Integrating AI and immersive VR into physical rehabilitation presents challenges that go beyond technology itself. A first limitation concerns the current evidence base. Although AI-driven immersive VR systems are promising, few clinical studies have specifically evaluated the combined use of AI and immersive VR in rehabilitation. Much of the available evidence comes from proof-of-concept systems, laboratory experiments, feasibility studies, or VR interventions with limited AI integration. Existing studies are also highly heterogeneous in terms of populations, technologies, outcomes, and intervention dosage. Many include small samples, short interventions, and limited long-term follow-up, making it difficult to assess sustained motor learning, adherence, functional transfer, and real-world clinical impact. Evidence also remains limited on whether AI-enhanced immersive VR offers added value over conventional rehabilitation, non-immersive digital tools, or standard VR interventions.
Another obstacle is the high financial cost of developing, implementing, and maintaining such systems. Many organizations struggle to absorb these expenses, even when the potential benefits include greater value, improved revenues, and increased efficiency (). These costs may be especially difficult to justify while evidence of long-term clinical benefit, cost-effectiveness, and scalability remains incomplete. Overcoming this barrier will require collaborative models that share costs across sectors, along with supportive regulatory frameworks that foster innovation while protecting patient interests.
Another significant challenge lies in end-user involvement and acceptance. The success of AI–VR rehabilitation systems depends heavily on continuous collaboration between developers, clinicians, and patients during the design, testing, and implementation phases to ensure that the technologies are clinically relevant, usable, and aligned with patient needs (). However, translating clinical rehabilitation needs into technical specifications can be challenging, particularly when communication between developers and end users is limited. Integrating VR technologies into routine clinical practice may also be difficult, as they must fit within existing workflows, documentation practices, and professional responsibilities while requiring ongoing training, technological confidence, and sustained engagement from healthcare professionals (Sadowski et al., 2026). Rather than immediately reducing workload, AI–VR systems may initially increase therapists’ responsibilities through training, calibration, supervision, interpretation of AI-generated outputs, troubleshooting, and safety monitoring. Some companies may help facilitate implementation by providing staff training, technical assistance, and troubleshooting support when technical issues arise (Sadowski et al., 2026). Encouraging interdisciplinary collaboration and maintaining flexible communication channels may help bridge these challenges and support the development of more intuitive and effective rehabilitation tools (Saghafian et al., 2021). This is especially important in rehabilitation, where person-centered care requires interventions to be tailored to patients’ capacities, goals, preferences, symptoms, and clinical context (; ). Digital literacy, technology confidence, language, sensory impairments, and socioeconomic factors may also influence patients’ ability to access and benefit from AI–VR rehabilitation, raising concerns about equity of access (Veras et al., 2023).
Sara Gerke and Cohen (2020) security concerns also arise with cloud-based platforms, as they can expose sensitive healthcare data to potential breaches; implementing robust cybersecurity measures such as encryption, multi-factor authentication, and isolated data networks is essential. These concerns are amplified when AI–VR systems continuously collect movement, behavioral, biometric, or physiological data, such as heart rate, muscle activity, eye movements, gait parameters, or performance metrics (). Data governance procedures must therefore address informed consent, data minimization, storage duration, secondary use of data, explainability of AI outputs, and the right of patients to understand how their data are used to guide rehabilitation decisions (). Likewise, ethical and legal issues surrounding AI-generated content (particularly regarding the use of copyrighted materials) require clear guidelines and compliance standards to ensure transparency, intellectual property protection, and accountability (Sara Gerke and Cohen, 2020). Finally, the environmental impact of AI, with its substantial energy demands, should not be overlooked, as underscored by recent studies ().
6 Conclusion
In conclusion, the synergy between AI and VR offers a powerful opportunity to address many current challenges in physical rehabilitation, while our review also highlights the specific limitations that arise when these technologies are combined. Through the creation of rich VR content and the optimization of sensory experiences, AI offers opportunities to enhance both the appeal and therapeutic potential of virtual environments. Rapid AI-driven adjustments to visual and auditory stimuli, tailored to users’ physiological and behavioral responses, have the potential to deepen immersion and clinical relevance of VR-based interventions. Applications such as BCIs for motor recovery, motion synthesis for realistic feedback, and the integration of flow and optimal-challenge principles are expanding the possibilities for rehabilitation delivery. Collectively, these innovations may contribute to more personalized, responsive, and effective therapies. As AI continues to evolve, its deeper integration with VR will likely bring more precise monitoring, adaptive therapeutic programs, and seamless clinician–patient collaboration across settings, transforming rehabilitation into a more engaging, efficient, and effective process for lasting recovery and independence.
Statements
Data availability statement
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding authors.
Author contributions
MB: Validation, Project administration, Visualization, Resources, Conceptualization, Writing – review and editing, Funding acquisition, Writing – original draft, Supervision. HD: Writing – review and editing, Visualization, Conceptualization, Project administration. DL: Writing – review and editing, Validation, Writing – original draft, Conceptualization. SH: Conceptualization, Writing – review and editing, Writing – original draft, Validation. MR: Writing – original draft, Validation, Conceptualization, Writing – review and editing. AL: Validation, Conceptualization, Writing – review and editing. SM: Conceptualization, Writing – review and editing, Validation. GL: Resources, Visualization, Funding acquisition, Conceptualization, Validation, Writing – review and editing.
Funding
The author(s) declared that financial support was received for this work and/or its publication. GL received salary support from the Fonds de recherche du Québec–Santé (FRQS).
Acknowledgments
The authors would like to express their gratitude to the Fondation FAiS, affiliated with the Centre intégré universitaire de santé et de services sociaux de la Capitale-Nationale, as well as to Mitacs for their support for the postdoctoral fellowship awarded to MB.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors SM, GL, AL, declared that they were an editorial board member of Frontiers at the time of submission. This had no impact on the peer review process and the final decision.
Generative AI statement
The author(s) declared that generative AI was used in the creation of this manuscript. The author(s) verify and take full responsibility for the use of generative AI in the preparation of this manuscript. ChatGPT was used solely to rephrase sentences and to improve the clarity, grammar, and quality of the English language. No generative AI tools were used to generate or interpret the scientific content, ideas, results, or conclusions presented in this work.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
1
AbediA.ColellaT. J. F.PakoshM.KhanS. S. (2024). Artificial intelligence-driven virtual rehabilitation for people living in the community: a scoping review. NPJ Digit. Med.7 (1), 25. 10.1038/s41746-024-00998-w
2
AdamovichS. V.FluetG. G.TunikE.MeriansA. S. (2009). Sensorimotor training in virtual reality: a review. NeuroRehabilitation25 (1), 29–44. 10.3233/NRE-2009-0497
3
Álvarez de la Campa CrespoM.DoneganT.Amestoy-AlonsoB.JustA.CombalíaA.Sanchez-VivesM. V. (2023). Virtual embodiment for improving range of motion in patients with movement-related shoulder pain: an experimental study. J. Orthop. Surg. Res.18 (1), 729. 10.1186/s13018-023-04158-w
4
AlvesT.GamaS.MeloF. S. (2018). Flow adaptation in serious games for health. 2018 IEEE 6th International Conference on Serious Games and Applications for Health (SeGAH). IEEE: Vienna, Austria.
5
AnantrasirichaiN.BullD. (2022). Artificial intelligence in the creative industries: a review. Artif. Intell. Rev.55 (1), 589–656. 10.1007/s10462-021-10039-7
6
BausO.BouchardS. (2014). Moving from virtual reality exposure-based therapy to augmented reality exposure-based therapy: a review. Front. Hum. Neurosci.8, 112. 10.3389/fnhum.2014.00112
7
BekeleM. K.PierdiccaR.FrontoniE.MalinverniE. S.GainJ. (2018). A survey of augmented, virtual, and mixed reality for cultural Heritage. J. Comput. Cult. Herit.11 (2), 1–36. 10.1145/3145534
8
BertonF.HoyetL.OlivierA. H.BruneauJ.MeurO. L.PettreJ. (2020). “Eye-Gaze activity in crowds: impact of virtual reality and density,” 2020 IEEE Conference on Virtual Reality and 3D User Interfaces (VR). 10.1109/TVCG.2020.2978971
9
BianY.YangC.GaoF.LiH.ZhouS.LiH.et al (2016). A framework for physiological indicators of flow in VR games: construction and preliminary evaluation. Personal Ubiquitous Comput.20, 1–12. 10.1007/s00779-016-0953-5
10
BordeleauM.StamenkovicA.TardifP. A.ThomasJ. (2022). The use of virtual reality in back pain rehabilitation: a systematic review and meta-analysis. J. Pain23 (2), 175–195. 10.1016/j.jpain.2021.08.001
11
BotellaC.Fernández-ÁlvarezJ.GuillénV.García-PalaciosA.BañosR. (2017). Recent progress in virtual reality exposure therapy for phobias: a systematic review. Curr. Psychiatry Rep.19 (7), 42. 10.1007/s11920-017-0788-4
12
CeradiniM.LosannoE.MiceraS.BandiniA.OrlandiS. (2024). Immersive VR for upper-extremity rehabilitation in patients with neurological disorders: a scoping review. J. Neuroeng Rehabil.21 (1), 75. 10.1186/s12984-024-01367-0
13
ChandlerD.MundayR. (2011). A Dictionary of Media and Communication. Oxford University Press.
14
ChenJ.OrC. K.ChenT. (2022). Effectiveness of using virtual reality-supported exercise therapy for upper extremity motor rehabilitation in patients with stroke: systematic review and meta-analysis of randomized controlled trials. J. Med. Internet Res.24 (6), e24111. 10.2196/24111
15
CipressoP.GiglioliI. A. C.RayaM. A.RivaG. (2018). The past, present, and future of virtual and augmented reality research: a network and cluster analysis of the literature. Front. Psychol.9, 2086. 10.3389/fpsyg.2018.02086
16
ClerwallC. (2014). Enter the robot journalist: users' perceptions of automated content. J. Practice8 (5), 519–531. 10.1080/17512786.2014.883116
17
CorreiaP. H. B. (2024). “Adaptive virtual reality solutions: a literature review,” in Proceedings of the 26th Symposium on Virtual and Augmented Reality. Manaus, Brazil: Association for Computing Machinery, 1–10.
18
CrawfordL.MaxwellJ.ColquhounH.KingsnorthS.FehlingsD.ZarshenasS.et al (2022). Facilitators and barriers to patient-centred goal-setting in rehabilitation: a scoping review. Clin. Rehabil.36 (12), 1694–1704. 10.1177/02692155221121006
19
CsikszentmihalyiM. (1988). Optimal Experience: Psychological Studies of Flow in Consciousness. Cambridge: Cambridge University Press.
20
DavidsJ.LidströmerN.AshrafianH. (2022). “Artificial intelligence for physiotherapy and rehabilitation,” in Artificial Intelligence in Medicine. Editors LidströmerN.AshrafianH. (Cham: Springer International Publishing), 1789–1807.
21
DemersM.FungK.SubramanianS. K.LemayM.RobertM. T. (2021). Integration of motor learning principles into virtual reality interventions for individuals with cerebral palsy: systematic review. JMIR Serious Games9 (2), e23822. 10.2196/23822
22
Dionne-DostieE.PaquetteN.LassondeM.GallagherA. (2015). Multisensory integration and child neurodevelopment. Brain Sci.5 (1), 32–57. 10.3390/brainsci5010032
23
DoneganT.Sanchez-VivesM. V. (2024). Perception and control of a virtual body in immersive virtual reality for rehabilitation. Curr. Opin. Neurol.37 (6), 638–644. 10.1097/wco.0000000000001321
24
FelnhoferA.PfannerstillF.GänslerL.KothgassnerO. D.HumerE.BüttnerJ.et al (2025). Barriers to adopting therapeutic virtual reality: the perspective of clinical psychologists and psychotherapists. Front. Psychiatry16, 1549090. 10.3389/fpsyt.2025.1549090
25
FisetF.LamontagneA.McFadyenB. J. (2023). Effects of age on the visuo-locomotor control used to circumvent a virtual pedestrian with different limb movements. Hum. Mov. Sci.87, 103026. 10.1016/j.humov.2022.103026
26
FraserA. D.BransonI.HollettR. C.SpeelmanC. P.RogersS. L. (2022). Expressiveness of real-time motion captured avatars influences perceived animation realism and perceived quality of social interaction in virtual reality. Front. Virtual Real.3, 981400. 10.3389/frvir.2022.981400
27
FrederiksenJ. G.SørensenS. M. D.KongeL.SvendsenM. B. S.Nobel-JørgensenM.BjerrumF.et al (2020). Cognitive load and performance in immersive virtual reality versus conventional virtual reality simulation training of laparoscopic surgery: a randomized trial. Surg. Endosc.34 (3), 1244–1252. 10.1007/s00464-019-06887-8
28
FreitasJ. R. S.VelosaV. H. S.AbreuL. T. N.JardimR. L.SantosJ. A. V.PeresB.et al (2021). Virtual reality exposure treatment in phobias: a systematic review. Psychiatr. Q.92 (4), 1685–1710. 10.1007/s11126-021-09935-6
29
FunkM.KoschT.SchmidtA. (2016). Interactive worker assistance: comparing the effects of in-situ projection, head-mounted displays, tablet, and paper instructions. UbiComp '16: Proceedings of the 2016 ACM International Joint Conference on Pervasive and Ubiquitous Computing. 934–939. 10.1145/2971648.2971706
30
GehrkeL.KoselevsA.KlugM.GramannK. (2025). Neuroadaptive haptics: a proof-of-concept comparing reinforcement learning from explicit ratings and neural signals for adaptive XR systems. Front. Virtual Real.6, 6. 10.3389/frvir.2025.1616442
31
GleggS. M. N.LevacD. E. (2018). Barriers, facilitators and interventions to support virtual reality implementation in rehabilitation: a scoping review. Pm R.10 (11), 1237–1251. 10.1016/j.pmrj.2018.07.004
32
Gonzalez-FrancoM.CohnB.OfekE.BurinD.MaselliA. (2020). The self-avatar follower effect in virtual Reality. IEEE Conference on Virtual Reality and 3D User Interfaces (VR).
33
GuenterB.FinchM.DruckerS.TanD.SnyderJ. (2012). Foveated 3D graphics. ACM Trans. Graph31 (6), 1–10. 10.1145/2366145.2366183
34
GuerraZ. F.LucchettiA. L. G.LucchettiG. (2017). Motor imagery training after stroke: a systematic review and meta-analysis of randomized controlled trials. J. Neurol. Phys. Ther.41 (4), 205–214. 10.1097/NPT.0000000000000200
35
GunnA. A. (1976). The diagnosis of acute abdominal pain with computer analysis. J. R. Coll. Surg. Edinb21 (3), 170–172.
36
GuoS.SouthernR.ChangJ.GreerD.ZhangJ. J. (2015). Adaptive motion synthesis for virtual characters: a survey. Vis. Comput.31 (5), 497–512. 10.1007/s00371-014-0943-4
37
Gutierrez-MartinezJ.Mercado-GutierrezJ. A.Carvajal-GámezB. E.Rosas-TriguerosJ. L.Contreras-MartinezA. E. (2021). Artificial intelligence algorithms in visual evoked potential-based brain-computer interfaces for motor rehabilitation applications: systematic review and future directions. Front. Hum. Neurosci.15, 772837. 10.3389/fnhum.2021.772837
38
HadlingtonL.ScaseM. O. (2018). End-user frustrations and failures in digital technology: exploring the role of fear of missing out, internet addiction and personality. Heliyon4 (11), e00872. 10.1016/j.heliyon.2018.e00872
39
HaimM.GraefeA. (2017). Automated news: better than expected?Digit. Journalism5 (8), 1044–1059. 10.1080/21670811.2017.1345643
40
IslamM.HuangT.AhnE.NaseemU. (2026). Multimodal generative AI for human motion understanding and generation: a survey and way forward. Inf. Fusion135, 104435. 10.1016/j.inffus.2026.104435
41
IssaT.IsaiasP. (2015). “User participation in the system development process,” in Sustainable Design: HCI, Usability and Environmental Concerns. Editors IssaT.IsaiasP. (London: Springer), 37–57.
42
JumaniA. K.ShiJ.LaghariA. A.EstrelaV. V.SampedroG. A.AlmadhorA.et al (2024). Quality of experience that matters in gaming graphics: how to blend image processing and virtual reality. Electronics13 (15), 2998. 10.3390/electronics13152998
43
KayesN. M.PapadimitriouC. (2023). Reflecting on challenges and opportunities for the practice of person-centred rehabilitation. Clin. Rehabil.37 (8), 1026–1040. 10.1177/02692155231152970
44
KemenyA.ChardonnetJ.-R.ColombetF. (2020). Getting Rid of Cybersickness: In Virtual Reality Augmented Reality and Simulators. Cham, Switzerland: Springer Nature Switzerland AG (Springer Cham).
45
KempitiyaT.SilvaD. D.RioE.SkarbezR.AlahakoonD. (2022). “Personalised physiotherapy rehabilitation using artificial intelligence and virtual reality gaming,” 2022 15th International Conference on Human System Interaction (HSI).
46
KiiliK. (2007). Foundation for problem-based gaming. Br. J. Educ. Technol.38 (3), 394–404. 10.1111/j.1467-8535.2007.00704.x
47
KilteniK.GrotenR.SlaterM. (2012). The sense of embodiment in virtual reality. Presence Teleoperators Virtual Environ.21 (4), 373–387. 10.1162/pres_a_00124
48
KornO.TietzS. (2017). “Strategies for playful design when gamifying rehabilitation: a study on user experience,” in Proceedings of the 10th International Conference on Pervasive Technologies Related to Assistive Environments. Island of Rhodes, Greece: Association for Computing Machinery, 209–214.
49
KritikosJ.AlevizopoulosG.KoutsourisD. (2021). Personalized virtual reality human-computer interaction for psychiatric and neurological illnesses: a dynamically adaptive virtual reality environment that changes according to real-time feedback from electrophysiological signal responses. Front. Hum. Neurosci.15, 596980. 10.3389/fnhum.2021.596980
50
KumaranV.RoweJ.MottB.LesterJ. (2023). SceneCraft: automating interactive narrative scene generation in digital games with large language models. Proc. AAAI Conf. Artif. Intell. Interact. Digital Entertain.19 (1), 86–96. 10.1609/aiide.v19i1.27504
51
LanotteF.O'BrienM. K.JayaramanA. (2023). AI in rehabilitation medicine: opportunities and challenges. Ann. Rehabil. Med.47 (6), 444–458. 10.5535/arm.23131
52
LedleyR. S.LustedL. B. (1959). Reasoning foundations of medical diagnosis. Science130 (3366), 9–21. 10.1126/science.130.3366.9
53
LevacD. E. (2024). Individual and contextual factors influencing children's effort in pediatric rehabilitation interventions. Dev. Med. Child Neurology66(1), 23-31. 10.1111/dmcn.15609
54
LiR.DuJ.YangK.WangX.WangW. (2022). Effectiveness of motor imagery for improving functional performance after total knee arthroplasty: a systematic review with meta-analysis. J. Orthop. Surg. Res.17 (1), 65. 10.1186/s13018-022-02946-4
55
LiH. H.LianJ. J.LiaoY. H. (2023). “Design an adaptive virtual reality game to promote elderly health,” 2023 International Conference on Computer, Information and Telecommunication Systems (CITS).
56
LiS.GaoC.ZhangJ.ZhangY.LiuY.GuJ.et al (2024). Less cybersickness, please: demystifying and detecting stereoscopic visual inconsistencies in virtual reality apps. Proc. ACM Softw. Eng.1 (FSE), 2167–2189. 10.1145/3660803
57
LiuL. Y.SanganiS.PattersonK. K.FungJ.LamontagneA. (2020). Real-time avatar-based feedback to enhance the symmetry of spatiotemporal parameters after stroke: instantaneous effects of different avatar views. IEEE Trans. Neural Syst. Rehabilitation Eng.28 (4), 878–887. 10.1109/TNSRE.2020.2979830
58
LiuX. J.GeS.CordovaA.YaghiZ.JiangB. Y.YueG. H.et al (2022). Elderly May benefit more from motor imagery training in gaining muscle strength than young adults: a systematic review and meta-analysis. Front. Psychol.13, 1052826. 10.3389/fpsyg.2022.1052826
59
LoH. H. M.ZhuM.ZouZ.WongC. L.LoS. H. S.ChungV. C.et al (2024). Immersive and nonimmersive virtual reality-assisted active training in chronic musculoskeletal pain: systematic review and meta-analysis. J. Med. Internet Res.26, e48787. 10.2196/48787
60
LuccioniS.JerniteY.StrubellE. (2024). Power Hungry Processing: Watts Driving the Cost of AI Deployment?Proceedings of the 2024 ACM conference on fairness, accountability, and transparency.
61
LustedL. B. (1955). Medical electronics. N. Engl. J. Med.252 (14), 580–585. 10.1056/nejm195504072521405
62
MacDormanK. F.ChattopadhyayD. (2016). Reducing consistency in human realism increases the uncanny valley effect; increasing category uncertainty does not. Cognition146, 190–205. 10.1016/j.cognition.2015.09.019
63
ChasséM. C.QuirionR.SiroisL.PinsonnaultS. (2022). “Inventer développer commercialiser sqri: stratégie québécoise de recherche et d’investissement en innovation 2022-2027. Editor l’Innovation MdlÉedBibliothèque et Archives nationales du Québec. Québec, QC: Gouvernement du Québec.
64
Martínez-CidS.HerreraV.Schez-SobrinoS.MonekossoD.Glez-MorcilloC.VallejoD. (2025). Artificial intelligence in physical therapy for neurological rehabilitation: a systematic mapping study. IEEE Access13, 1–185081. 10.1109/access.2025.3626013
65
Matamala-GomezM.MalighettiC.CipressoP.PedroliE.RealdonO.MantovaniF.et al (2020). Changing body representation through full body ownership illusions might foster motor rehabilitation outcome in patients with stroke. Front. Psychol.11, 1962. 10.3389/fpsyg.2020.01962
66
MennellaC.ManiscalcoU.PietroG. D.EspositoM. (2023). The role of artificial intelligence in future rehabilitation services: a systematic literature review. IEEE Access11, 11024–11043. 10.1109/access.2023.3236084
67
MillerL.ZivianiJ.WareR. S.BoydR. N. (2016). Does context matter? Mastery motivation and therapy engagement of children with cerebral palsy. Phys. Occup. Ther. Pediatr.36 (2), 155–170. 10.3109/01942638.2015.1076556
68
MoghimiM.StoneR.RotshteinP.CookeN. (2016). Influencing human affective responses to dynamic virtual environments. Presence Teleoper Virtual Environ.25 (2), 81–107. 10.1162/pres_a_00249
69
MunzertJ.LoreyB.ZentgrafK. (2009). Cognitive motor processes: the role of motor imagery in the study of motor representations. Brain Res. Rev.60 (2), 306–326. 10.1016/j.brainresrev.2008.12.024
70
OlulekeB.XumingY. (2013). “Virtual reality and virtual reality system components,” Proceedings of the 2nd International Conference on Systems Engineering and Modeling (ICSEM 2013) (Paris, France: Atlantis Press).
71
O'SheaH.MoranA. (2017). Does motor simulation theory explain the cognitive mechanisms underlying motor imagery? A critical review. Front. Hum. Neurosci.11, 72. 10.3389/fnhum.2017.00072
72
ParkM.LeeM.YuS. (2022). HRGAN: a generative adversarial network producing higher-resolution images than training sets. Sensors22 (4), 1435. 10.3390/s22041435
73
PatneyA.SalviM.KimJ.KaplanyanA.WymanC.BentyN.et al (2016). Towards foveated rendering for gaze-tracked virtual reality. ACM Trans. Graph35 (6), 1–12. 10.1145/2980179.2980246
74
PetersR.GraeffA.PaulC. (1997). “Integrating agents into virtual worlds,” in Proceedings of the 1997 Workshop on New Paradigms in Information Visualization and Manipulation; Las Vegas. Nevada, USA: Association for Computing Machinery, 69–74.
75
PetsaniD.KonstantinidisE.KatsouliA.-M.ZilidouV.DiasS. B.HadjileontiadisL.et al (2022). Digital biomarkers for well-being through exergame interactions: exploratory study. JMIR Serious Games10 (3), e34768. 10.2196/34768
76
PlavoukouT.StaktopoulosP.PapagiannisG.StasinopoulosD.GeorgoudisG. (2025). Virtual and augmented reality for chronic musculoskeletal rehabilitation: a systematic review and exploratory meta-analysis. Bioengineering12 (7), 745. 10.3390/bioengineering12070745
77
PoulsenA. A.RodgerS.ZivianiJ. M. (2006). Understanding children's motivation from a self-determination theoretical perspective: implications for practice. Aust. Occup. Ther. J.53 (2), 78–86. 10.1111/j.1440-1630.2006.00569.x
78
RivaG.MantovaniF.GaggioliA. (2004). Presence and rehabilitation: toward second-generation virtual reality applications in neuropsychology. J. Neuroeng Rehabil.1 (1), 9. 10.1186/1743-0003-1-9
79
RulleauT.ToussaintL. (2014). L’imagerie motrice en rééducation. Kinésithérapie, Rev.14 (148), 51–54. 10.1016/j.kine.2013.12.015
80
RutkowskiS.KiperP.CaccianteL.CieślikB.MazurekJ.TurollaA.et al (2020). Use of virtual reality-based training in different fields of rehabilitation: a systematic review and meta-analysis. J. Rehabil. Med.52 (11), jrm00121. 10.2340/16501977-2755
81
SadowskiI.Gagnon-MailhotM.SeuG. J.BatchoC. S.LabergeM.KhouryB.et al (2026). Barriers and facilitators to implementing immersive virtual reality in long-term care settings: an interdisciplinary partnership study exploring staff perspectives. Front. Pain Res. (Lausanne)7, 1734386. 10.3389/fpain.2026.1734386
82
SaghafianM.LaumannK.SkogstadM. R. (2021). Organizational challenges of development and implementation of virtual reality solution for industrial operation. Front. Psychol.12, 704723. 10.3389/fpsyg.2021.704723
83
SaidR. R.HeyatM. B. B.SongK.TianC.WuZ. (2022). A systematic review of virtual reality and robot therapy as recent rehabilitation technologies using EEG-Brain-Computer interface based on movement-related cortical potentials. Biosens. (Basel).12 (12), 1134. 10.3390/bios12121134
84
Sara GerkeT. M.CohenG. (2020). Ethical and legal challenges of artificial intelligence-driven healthcare. Artif. Intell. Healthc.295–336. 10.1016/B978-0-12-818438-7.00012-5
85
SchindlerS.ZellE.BotschM.KisslerJ. (2017). Differential effects of face-realism and emotion on event-related brain potentials and their implications for the uncanny valley theory. Sci. Rep.7, 45003. 10.1038/srep45003
86
SchraderC.BrichJ.FrommelJ.RiemerV.RogersK. (2017). “Rising to the challenge: an emotion-driven approach toward adaptive serious games,” in Serious Games and Edutainment Applications. Editors MaM.OikonomouA. (Cham: Springer International Publishing), 3–28. 10.1007/978-3-319-51645-5_1
87
ShahS. H. H.KarlsenA. S. T.SolbergM.HameedI. A. (2023). A social VR-based collaborative exergame for rehabilitation: codesign, development and user study. Virtual Real.27 (4), 3403–3420. 10.1007/s10055-022-00721-8
88
SikkaR.SaxenaV.NachappaM. (2024). “An innovation of exploiting generative adversarial networks to augment virtual reality experiences,” 2024 International Conference on Optimization Computing and Wireless Communication (ICOCWC). IEEE: Debre Tabor, Ethiopia. 10.1109/ICOCWC60930.2024.10470680
89
SingareddyS.SnV. P.JaramilloA. P.YasirM.IyerN.HusseinS.et al (2023). Artificial intelligence and its role in the management of chronic medical conditions: a systematic review. Cureus15 (9), e46066. 10.7759/cureus.46066
90
SingerT.FaheyP.LiuK. P. Y. (2024). Effectiveness of motor imagery in the rehabilitation of people with parkinson's disease: a systematic review and meta-analysis. Neurorehabil Neural Repair38 (6), 460–475. 10.1177/15459683241246493
91
SlaterM.LottoB.ArnoldM. M.Sanchez-VivesM. V. (2009). How we experience immersive virtual environments: the concept of presence and its measurement. Anu. Psicol.40 (2), 193–210. 10.1145/3772318.3790724
92
SuhA.ProphetJ. (2018). The state of immersive technology research: a literature analysis. Comput. Hum. Behavior86, 77–90. 10.1016/j.chb.2018.04.019
93
SumnerJ.LimH. W.ChongL. S.BundeleA.MukhopadhyayA.KayambuG. (2023). Artificial intelligence in physical rehabilitation: a systematic review. Artif. Intell. Med.146, 102693. 10.1016/j.artmed.2023.102693
94
Teslasuit (2025). Teslasuit Dev Kit. London, United Kingdom: Deep Divers Ltd (Teslasuit). Available online at: https://teslasuit.io/products/teslasuit-4/ (Accessed March 15, 2025).
95
VenturaS.MarchettiP.BañosR.TessariA. (2023). Body ownership illusion through virtual reality as modulator variable for limbs rehabilitation after stroke: a systematic review. Virtual Real.27 (3), 2481–2492. 10.1007/s10055-023-00820-0
96
VerasM.LabbéD. R.FurlanoJ.ZakusD.RutherfordD.PendergastB.et al (2023). A framework for equitable virtual rehabilitation in the metaverse era: challenges and opportunities. Front. Rehabil. Sci.4, 1241020. 10.3389/fresc.2023.1241020
97
VerschuerenS.BuffelC.Vander SticheleG. (2019). Developing theory-driven, evidence-based serious games for health: framework based on research community insights. JMIR Serious Games7 (2), e11565. 10.2196/11565
98
Villegas-ChW.Maldonado NavarroA.Mera-NavarreteA. (2025). Using generative adversarial networks for the synthesis of emotional facial expressions in virtual educational environments. Intelligent Syst. Appl.25, 200479. 10.1016/j.iswa.2025.200479
99
WangS.SunJ.YinX.LiH. (2023). Effect of virtual reality technology as intervention for people with kinesiophobia: a meta-analysis of randomised controlled trials. J. Clin. Nurs.32 (13-14), 3074–3086. 10.1111/jocn.16397
100
WeiW.McElroyC.DeyS. (2019). Towards On-Demand virtual physical therapist: machine learning-based patient action understanding, assessment and task recommendation. IEEE Trans. Neural Syst. Rehabil. Eng.27 (9), 1824–1835. 10.1109/TNSRE.2019.2934097
101
WenkN.Penalver-AndresJ.BuetlerK. A.NefT.MüriR. M.Marchal-CrespoL. (2023). Effect of immersive visualization technologies on cognitive load, motivation, usability, and embodiment. Virtual Real27 (1), 307–331. 10.1007/s10055-021-00565-8
102
WinklerA.WonJ.YeY. (2022). “QuestSim: human motion tracking from sparse sensors with simulated avatars,” in SIGGRAPH Asia 2022 Conference Papers. Daegu, Republic of Korea. New York, NY:Association for Computing Machinery.
103
YalcinM.HalbigA.FischbachM.LatoschikM. E. (2024). Automatic cybersickness detection by deep learning of augmented physiological data from off-the-shelf consumer-grade sensors. Front. Virtual Real.5. 10.3389/frvir.2024.1364207
104
YanT.LiangW.ChanC. W. H.ShenY.LiuS.LiM. (2025). Effects of motor imagery training on gait performance in individuals after stroke: a systematic review and meta-analysis. Disabil. Rehabil.47 (1), 47–61. 10.1080/09638288.2024.2337091
105
YannakakisG. N.HallamJ. (2009). Real-time game adaptation for optimizing player satisfaction. IEEE Trans. Comput. Intell. AI Games1 (2), 121–133. 10.1109/tciaig.2009.2024533
106
YeG.KohR. G. L.JaiswalK.SoomalH.KumbhareD. (2023). The use of virtual reality in the rehabilitation of chronic nonspecific neck pain: a systematic review and meta-analysis. Clin. J. Pain39 (9), 491–500. 10.1097/ajp.0000000000001134
Summary
Keywords
human–machine interaction, machine learning, patient engagement, personalized rehabilitation, real-time adaptation
Citation
Bordeleau M, Derue H, Labbé DR, Houzangbe S, Robert MT, Lamontagne A, Marchand S and Léonard G (2026) Next-generation movement-based physical rehabilitation through artificial intelligence-driven immersive virtual reality: a brief perspective. Front. Virtual Real. 7:1760132. doi: 10.3389/frvir.2026.1760132
Received
03 December 2025
Revised
04 July 2026
Accepted
09 July 2026
Published
10 August 2026
Volume
7 - 2026
Edited by
Silvia Orlandi, University of Bologna, Italy
Reviewed by
Tony Donegan, August Pi i Sunyer Biomedical Research Institute (IDIBAPS), Spain
Updates
Copyright
© 2026 Bordeleau, Derue, Labbé, Houzangbe, Robert, Lamontagne, Marchand and Léonard.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Martine Bordeleau, martine.bordeleau@outlook.com; Guillaume Léonard, guillaume.leonard2@usherbrooke.ca
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.