What can virtual-reality arm training do after stroke?

The short answer

During rehabilitation, adults recovering from stroke who practised arm movements with an interactive screen or virtual-reality program showed better broad arm movement and wider joint range in the affected arm than people receiving usual rehabilitation or no specific therapy. Measures of everyday independence also improved in some analyses, but there was no clear overall benefit for fine finger control, grip strength, muscle tightness, quality of life, or how much the affected arm was actually used in daily routines. Most trials stopped measuring when training ended, so it is uncertain whether these gains remain after the program stops; the limited follow-up evidence did not show clear maintenance, but that does not prove the benefits disappeared.

What the virtual-reality arm training actually involved

The training was not one standard program. In most of the 43 randomized trials—studies where people were assigned to treatments by chance—participants used a screen-based, non-immersive setup, often something like a Nintendo Wii, rather than putting on a headset. They usually held a controller that responded to broad arm and hand movements: picking it up, reaching, lifting and moving it. Some trials used gloves with sensors or a camera that tracked the hand, such as Leap Motion, which allowed finer finger exercises. About two-thirds of trials used programs designed specifically for rehabilitation, and about one-third used commercial games. Typical sessions lasted 20 to 75 minutes, several times a week, over roughly two weeks to two months.

The comparison groups make the meaning clearer

The trials compared virtual-reality exercise against what people would otherwise receive. In most trials (28 of 43), virtual-reality exercise was added on top of conventional rehabilitation, and the comparison group received the same conventional rehabilitation without the extra virtual reality. This answer tells us whether adding virtual reality to usual care gives extra benefit. A smaller set (13 trials) compared virtual-reality exercise alone with conventional rehabilitation, which tests whether it might replace usual therapy. Only two trials compared virtual reality alone with no therapy. So the strong findings mostly reflect extra screen-based practice on top of usual therapy, not a replacement for it.

What improved during the training period

When outcomes were measured at the end of the scheduled therapy, people assigned to virtual-reality training had, on average, better scores on a standard test of how well the affected arm moves—the Fugl-Meyer upper extremity score. The advantage was around half a standard deviation. Range of motion at the shoulder, elbow or wrist, measured with a goniometer, was about one standard deviation better. Muscle strength assessed by a therapist also favored the virtual-reality group in one small set of trials. Independence in everyday activities improved on two common assessment tools: one that rates help needed with self-care and mobility, and another that scores overall disability after stroke. These improvements were found at the end of the scheduled therapy program.

What did not change, and why fine finger control lagged

The combined evidence did not show a reliable overall improvement in fine finger control, measured by a test that involves picking up small objects and placing them precisely. Overall hand dexterity, grip strength measured with a squeezing device, muscle tightness called spasticity, and quality-of-life ratings were also similar between the virtual-reality and comparison groups. The authors suggested a practical reason for the lack of fine-finger benefit: most systems used handheld controllers that mainly require grasping and moving the whole controller, so the exercises did not demand the delicate, separate finger movements that everyday tasks such as fastening a button require. Systems using sensor gloves or hand-tracking cameras may be better suited to that kind of training, but that possibility needs direct testing.

Everyday independence and actual arm use are not the same measurement

The review found improved scores on tools that ask how independently someone performs daily activities, such as eating, bathing, dressing and moving between positions. That is different from measuring whether the affected arm is actually chosen spontaneously during the day. On a self-report log of real-world arm use, no clear improvement emerged. In other words, people may have become more capable of using the arm when needed during structured rehabilitation, but the evidence did not show them automatically using it more in everyday routines outside that setting.

Patterns from longer or added training—read cautiously

When the researchers split the trials into smaller groups, some patterns appeared. Hand dexterity seemed to improve more when people received more than 15 hours of virtual-reality practice compared with 15 hours or fewer, when the program ran for more than one month, and when virtual reality was added to conventional therapy rather than used alone. These are exploratory comparisons from splits that can easily produce chance findings, and the authors cautioned that they are less reliable than the main results. Still, the pattern is consistent with the simple idea that more total practice time is linked to better arm recovery. In the main analysis of broad arm movement, both purpose-built rehabilitation programs and commercial games showed benefit, and both younger and older participants improved.

What follow-up can and cannot tell us

Only 14 of the 43 trials arranged any assessment after the training program ended; 29 did not. In the follow-up data that were available, the authors found no evidence that the improvements were maintained after the intervention stopped, and almost all outcomes were no longer significantly different between groups. The exception was a small signal on one test: people in the virtual-reality groups completed everyday arm tasks slightly faster when measured from the start of the study to later follow-up. This does not prove that the gains disappeared, because the trials did not report what rehabilitation or exercises participants received during the follow-up period; after-training activities could have been similar in both groups or changed in ways that washed out differences. The honest conclusion is that persistence remains unknown.

What this evidence adds

This systematic review, which combined results across 43 randomized trials with about 1,900 people, shows that screen-based virtual-reality arm practice can improve broad arm movement and joint range during rehabilitation, and can support gains in everyday independence. It also shows that this kind of training had not clearly improved fine finger control, and that most trials did not follow people long enough to know whether the benefits last. Many of the trials were small, and the programs varied, so range-of-motion and muscle strength findings should be interpreted cautiously. The practical message is that virtual-reality practice appears useful as an added way to increase guided arm practice during post-stroke rehabilitation, not as a proven replacement for conventional therapy or a guaranteed way to improve fine hand skills.