What is the potential of virtual reality for post-stroke sensorimotor rehabilitation?

KEYWORDS:

• Rehabilitation
• recovery
• computer-assisted
• stroke
• motor control
• motor learning
• neuroplasticity

1. Origin and state-of-the-art

Simulated environments have been developing since they were first introduced by Ivan Sutherland in 1965 [1], but it was not until the middle to late 1990s that virtual reality applications began to emerge as tools for cognitive and motor rehabilitation. Since that time, technological advances have outpaced scientific validation of the approach such that by the time evidence of the effectiveness of a specific technology has accumulated, the technology may already have become obsolete. Nevertheless, both technology and evidence are continuing to advance. Technological advances include higher resolution and less encumbering head-mounted displays and motion tracking systems. For example, whereas previous motion tracking was mainly done with wired markers placed on different body segments that confined the user to move within a fixed environment, the accuracy and resolution of markerless, portable systems is advancing. The use of low-cost, higher resolution head-mounted displays is also making visually immersive virtual applications more accessible. However, head-mounted displays still have limitations in their field of view affecting the quality of the viewing environment, which may have implications for the formation of realistic movement patterns. Technological advances also include embedded ambient technologies capable of rendering text, graphics, and sound in relation to user actions and of monitoring real-world activity from a distance so that therapy can be accessible to those unable to travel to a clinic or treatment center. Augmented reality applications are advancing in which a range of elements such as visual graphics and sounds can be superimposed from a computer screen onto a real time environment. One advantage of augmented reality is the provision of the user with haptic feedback about object properties that is not available in virtual environments without the use of end effector accoutrements.

Virtual reality for rehabilitation is the art and science of creating and applying interactive cognitive and/or physical activities created with computers that appear and feel similar to real-world objects and events [2]. Virtual rehabilitation goes beyond interacting with objects in the physical world by incorporating activities that may be too challenging or hazardous if practiced in reality (i.e., facing ones phobias, walking on slippery surfaces that may lead to a fall) and creating interactive practice environments that are motivating and fun.

Virtual rehabilitation for sensorimotor rehabilitation in people with neurological disorders offers several other advantages over standard conventional therapy. To maximize sensorimotor recovery after brain lesions, physical training needs to be long-lasting, challenging, intensive, salient, and motivating [3]. Animal studies suggest that 400–600 repetitions per day of functional tasks are required to induce structural neurological changes [4]. Virtual training environments provide such opportunities to engage in intensive training that may not be available to the individual due to constraints in health-care delivery systems. Constraints such as limited time, personnel, or resources, affect the amount of therapy that can be delivered in inpatient and outpatient rehabilitation settings. For example, Lang et al. [5] found that patients in an outpatient stroke unit practiced an average of only 41 active and purposeful movement repetitions of the upper limb per session, far below the number needed for motor skill acquisition and the recommended dose according to national and international stroke rehabilitation guidelines [6]. To address this problem, VR training platforms may be used as adjunctive therapy in which opportunities for extended practice may be made available for patients with supervision in groups or with specialized support personnel [7]. Nevertheless, challenges to the implementation of VR applications also need to be addressed. Some barriers that have been identified as limiting the uptake of virtual technology by clinicians is a lack of time and expertise to learn how to use the technology, a lack of ability of some applications to be modified to address specific patient goals and distraction from other rehabilitation goals [8].

Aside from intensity, for sensorimotor rehabilitation, challenge level, and feedback that is salient to the individual and/or the task are essential factors for experience-dependent neuroplasticity [3] that can be exploited in virtual rehabilitation applications. These elements of training are important to optimize motor recovery by challenging the system to engage in adaptive motor strategies to find solutions to different motor problems [9–11]. This emergent rather than a reductive approach to motor relearning draws on the general idea that skill learning is reflected in the mastery of redundant degrees of freedom [9]. The physical rather than biomechanical approach to motor control acknowledges that movement planning involves problem-solving based on available constraints and possibilities for action, including those related to the organism (i.e., cognitive, biomechanical), the environment, and the task [10]. Thus, the opportunity to engage in high-intensity varied practice, distinguishes virtual rehabilitation from technology-based approaches advocating single-task repetitive training such as some of those employing passive or active robotic devices (e.g. [12]).

Challenge level and feedback are also important to sensorimotor recovery since they enhance patient engagement and motivation. In particular, social-cognitive and affective motivation associated with engaging in different practice scenarios [13] combined with the provision of feedback [14 15,] enabled by virtual rehabilitation may positively influence motor learning. Indeed, positive feedback signaling reward engages the mesolimbic dopaminergic system, which has been associated with working memory and motor system functions such as planning and response selection [16]. Activities that are interesting and engaging also engender a desire by the individual to practice longer and/or harder, promoting experience-dependent neural plasticity [3].

2. Does virtual rehabilitation deliver these practice environments?

Evidence based on randomized controlled trials has reached the level where recommendations about the effectiveness of virtual rehabilitation applications can be made [17, 18]. However, this evidence remains at a low-quality level due to the risk of bias and inconsistencies between studies. Based on a meta-analysis of 22 studies, including 1038 participants with stroke, Laver et al. [17] found that compared to conventional therapy, virtual rehabilitation was not superior for improving upper limb function at an equal intensity of practice for post-acute patients with stroke. However, when virtual rehabilitation was used to supplement usual care so that participants in the intervention group received a higher dose of therapy (10 studies, 210 participants), greater improvements in upper limb function occurred. Benefits were greater for virtual reality interventions specifically designed for rehabilitation than those using off‐the‐shelf gaming programs. For lower limb functions of balance and gait in patients with mostly post-acute stroke, de Rooij et al. (18) found that virtual rehabilitation was more effective when added to conventional therapy and dose-matched for time in therapy, compared to conventional treatment alone. Thus, virtual rehabilitation as an adjunctive therapy has the benefit of increasing the amount of therapy delivered to the patient, which is linked to better outcomes, and has made its way into clinical practice guidelines for stroke care [6]. However, there is still a lack of evidence of the effectiveness of virtual reality applications for cognitive post-stroke rehabilitation and on participation and activities of daily living [17].

3. How can virtual rehabilitation better meet the needs of sensorimotor rehabilitation?

While training elements important for motor learning (i.e., how to practice: intensity, challenge point) are being integrated into virtual rehabilitation systems, the incorporation of motor control principles (i.e., what types of motor actions to practice) has been less evident. Applications often do not take into account basic elements of movement production such as object affordances and reachability (i.e., environments calibrated to an individual’s actual arm length), which are essential for perception-action coupling [19]. Visual environments are often represented in two instead of three dimensions making judgments of depth perception unreliable. Another general problem in motor retraining is that task practice focusing on only endpoint (i.e., hand or foot) performance variables can be equally accomplished with desirable or undesirable (i.e., compensatory) movements of different joints or segments of the moving limb and trunk. This often is compounded by the provision of feedback only about whether the task was successfully accomplished and not about how the task was done. It is important to remember that based on the system’s redundancy, the same movement can be accomplished by a variety of joint rotations, and that the system will take advantage of its redundancy to solve the motor problem by finding the easiest possible solution. This often leads to the appearance of compensatory movements being incorporated into a task, such as excessive forward trunk displacement while reaching for a cup or other object [20]. Replacement of typical movement patterns with compensatory movements results in ‘bad plasticity’ or changes in neural connectivity that may reinforce undesirable movement patterns [21]. Once learned, replacement of the compensatory pattern with the desirable one, is difficult, and may represent a barrier to recovery [21]. Thus, a lack of emphasis on and feedback about movement quality combined with poor limb motion tracking in virtual environments using low-cost camera-based trackers, may actually increase the possibility of individuals intensively practicing non-desirable movements.

4. Five-year view: where do we go from here?

Consensus opinion is that the advantage of virtual rehabilitation is in the possibility of increasing time in therapy by providing adjunctive or additional therapy sessions for both the upper and lower limbs [17, 18]. It is also generally agreed that virtual rehabilitation can increase the motivation of the individual to practice more often and/or more intensively because of engaging scenarios and the prospect of positive rewards. One of the remaining challenges for the virtual rehabilitation community is to design training activities that are more strongly based on principles of motor control and motor learning than those currently available. Applications can include challenging activities related to the skill level of the learner and provide salient feedback aimed at maximizing motor problem-solving, rather than repetition of a prescribed movement or set of movements.

It is also now increasingly being recognized that no single physical intervention is superior to any other [22] on improving post-stroke upper limb recovery based on disappointing results from several large scale randomized controlled trials on task-specific training and robotics (e.g. [12]). Further studies may emphasize the delivery of a combination of therapies, focusing on the individual needs and goals of the learner. Virtual rehabilitation interventions can be included as an adjunctive treatment modality within the context of programs of individualized patient care, both for the upper- and the lower-limb. Future large-scale randomized controlled trials are needed in which individualized practice regimens can be combined with different technologies such as virtual reality, robotics, and electrical stimulation for maximal rehabilitation effectiveness.

Declaration of interest

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Additional information

Funding

This paper was not funded.