Robotic rehabilitation has moved from engineering laboratories into stroke rehabilitation clinics around the world.
Some devices support the shoulder and elbow. Others guide the hand or wrist. Exoskeletons move multiple joints, while end-effector robots interact with the hand or forearm.
Their promise is attractive: a robot can help a patient perform hundreds of highly repetitive movements during a single session, while measuring performance and adjusting assistance.
But an important question remains:
Does robotic therapy actually produce better recovery than conventional rehabilitation?
The current evidence suggests a nuanced answer.
Robotic therapy can improve upper-extremity motor impairment after stroke, but the size of the additional benefit depends on the patient, timing, device, and amount of therapy being compared.
Why Robots Are Attractive for Stroke Rehabilitation
Upper-extremity rehabilitation requires considerable practice.
Robotic systems can make high-volume training easier by providing:
Repetitive movement practice
Adjustable physical assistance
Resistance when appropriate
Arm-weight support
Immediate performance feedback
Quantitative movement measurements
A large 2024 meta-analysis in Neurology included 90 randomized controlled trials involving 4,311 stroke survivors. Robotic programs averaged roughly 806 repetitions per session in studies that reported repetition counts, compared with about 358 in control groups.
That capacity to deliver many repetitions is one of robotics' clearest practical advantages.
Do Robots Improve Fugl-Meyer Scores?
The answer is generally yes—but the average benefit may be modest.
The 2024 Neurology meta-analysis found that robotic rehabilitation improved upper-extremity Fugl-Meyer scores by an average of 2.23 points more than non-robotic therapy. The researchers described this as a statistically significant but small improvement and found that the motor-impairment gains did not consistently translate into better upper-limb functional capacity.
This distinction matters.
Improving a Fugl-Meyer score and becoming better at tasks such as drinking from a cup, dressing, or preparing food are related outcomes—but they are not identical.
What Happens When Therapy Dose Is Matched?
One of the biggest questions in robotic rehabilitation is whether improvement occurs because of the robot itself or simply because the robot allows patients to receive more therapy.
A major 2025 systematic review addressed this by examining 54 studies with 2,744 participants in which robotic rehabilitation was compared with dose-matched conventional treatment.
Robotics produced a small advantage in upper-limb capacity:
SMD 0.14
However, the advantage disappeared at follow-up, and there was no significant improvement in activities of daily living compared with conventional therapy of similar dose. The authors concluded that the overall difference was not clinically meaningful.
This finding is important:
Robotic therapy may be very useful for delivering intensive rehabilitation, but the robot itself is not necessarily superior to an equally intensive, well-designed therapist-led program.
Some Patients May Benefit More Than Others
Recent evidence suggests that averaging all stroke survivors together may hide important differences.
A 2026 systematic review and meta-analysis examined 42 randomized trials involving 1,678 participants and analyzed outcomes according to recovery stage and impairment severity.
The researchers reported particularly large benefits among subacute patients and patients with severe impairment. In these groups, robotic therapy was associated with an approximately 8.82-point greater improvement in FMA-UE, along with improvements in activities of daily living.
That result is substantially larger than the average effect reported when all patients and robot types are pooled together.
It suggests that the question may eventually become less:
“Does robotic therapy work?”
and more:
“Which patient should receive which robotic intervention, and when?”
Timing May Matter
The early weeks and months after stroke represent an important period of neurological recovery.
The 2024 Neurology meta-analysis found evidence suggesting that patients with remaining recovery potential may benefit more from robotic treatment, particularly during the acute and early-subacute periods.
A more recent multicenter randomized trial of exoskeleton-assisted treatment in early-subacute stroke also reported a substantially higher proportion of patients reaching a prespecified clinically important Fugl-Meyer improvement with robotic therapy than with the comparison intervention.
However, individual trials should not be interpreted as proof that every early-stroke patient should receive a robot. Patient selection and replication remain important.
The Type of Robot Matters
“Robotic rehabilitation” is not one treatment.
Systems differ considerably.
Exoskeleton Robots
An exoskeleton aligns mechanical joints with the patient's shoulder, elbow, wrist, or hand.
These systems can provide:
Arm-weight support
Multi-joint assistance
Controlled movement trajectories
The 2024 Neurology analysis found a small significant benefit in upper-limb capacity specifically for exoskeleton devices, while that benefit was not observed across robotics as a whole.
End-Effector Robots
These devices typically connect to the hand or forearm and guide movement through an external contact point.
A 2025 network meta-analysis involving 31 randomized trials found improvements in FMA-UE when robot-assisted therapy was added to conventional rehabilitation and suggested differences between end-effector and exoskeleton systems depending on the outcome being considered.
There is therefore no scientifically justified reason to assume that all rehabilitation robots produce the same effect.
What About the Hand?
This remains one of the major challenges.
Many upper-extremity robots emphasize shoulder and elbow movement because these joints are mechanically easier to assist.
Recovering:
Finger individuation
Grasp
Release
Fine manipulation
Thumb control
is considerably more complicated.
A patient's proximal arm movement may therefore improve without an equivalent improvement in hand function.
For stroke survivors whose primary limitation is hand use, it is important to determine which joints the device actually trains before assuming that an “upper-extremity robot” addresses the patient's main impairment.
Does Robotic Therapy Reduce Spasticity?
Evidence is much less convincing.
The large 2024 Neurology meta-analysis found no significant overall effect on muscle tone compared with non-robotic interventions.
Robotic therapy should therefore not be presented as a general treatment for spasticity.
Spasticity management may require a different combination of rehabilitation, positioning, stretching, medication, injections, or other interventions based on individual clinical assessment.
The Biggest Advantage May Be Therapy Dose
Robots have an important characteristic that is sometimes overlooked:
They do not get tired of repetitions.
A therapist may need to divide attention among multiple goals during a session. A robotic device can potentially supervise hundreds of targeted movements while continuously measuring performance.
This creates a compelling model in which robots extend clinicians' capacity rather than replace clinicians.
For example:
A clinician assesses impairment.
The clinician chooses appropriate movement targets.
The robot provides high-repetition training.
Movement data are collected automatically.
The clinician reviews progress and modifies treatment.
The therapeutic value may therefore come from combining human clinical reasoning with scalable technology.
Robots Should Not Replace Functional Practice
A patient ultimately needs to use the arm outside the robot.
Drinking from a cup, buttoning a shirt, preparing a meal, opening a door, or carrying groceries requires movement in unpredictable real-world environments.
The 2025 dose-matched meta-analysis found no meaningful overall advantage for robotics in activities of daily living, despite some improvement in upper-limb capacity.
For this reason, robotic training is best viewed as one component of rehabilitation, alongside:
Task-specific practice
Functional activities
Strength and mobility training
Hand training
Home rehabilitation
Therapist feedback
Robotics and Recovery Tracking
Robotic systems also have another potentially valuable role: measurement.
During each movement, a robot can potentially record:
Range of motion
Speed
Accuracy
Force
Number of repetitions
Assistance required
Movement trajectory
Tracking these measures across weeks could reveal changes that are difficult to detect visually during an occasional clinic appointment.
Combining robotic measurements with standardized clinical outcomes such as the FMA-UE may eventually provide clinicians with a richer picture of recovery.
But these device-generated measures require proper validation before they should be interpreted as clinical outcomes.
What Should Stroke Survivors Ask Before Starting Robotic Therapy?
Rather than asking only whether a clinic has a rehabilitation robot, useful questions include:
Which joints does the device train?
Is the device appropriate for my impairment level?
How many repetitions will I perform?
Will robotic training supplement or replace my current therapy?
How will my progress be measured?
Will I also practice meaningful functional activities?
Is the treatment targeting my hand, proximal arm, or both?
The answers may matter more than the brand of robot.
Final Thoughts
Robotic therapy is neither a miracle treatment nor an ineffective technological novelty.
The strongest interpretation of current evidence is somewhere between those extremes.
Robots can provide intensive, repeatable, measurable upper-extremity training, and several contemporary studies show improvements in motor impairment after stroke. Some evidence suggests particularly meaningful benefits for selected patients during the subacute stage and for those with severe impairment.
At the same time, large meta-analyses show that when robotic and conventional therapy are carefully matched for treatment dose, the average additional benefit of robotics may be small, particularly for real-world arm function and activities of daily living.
The future of robotic rehabilitation may therefore depend less on building a robot that replaces therapy and more on building systems that help clinicians deliver more personalized, intensive, measurable rehabilitation to the patients most likely to benefit.
References
De Iaco L, Veerbeek JM, Ket JCF, Kwakkel G. Upper Limb Robots for Recovery of Motor Arm Function in Patients With Stroke: A Systematic Review and Meta-Analysis. Neurology. 2024;103(2):e209495. Ninety RCTs and 4,311 participants were included; average FMA-UL benefit was 2.23 points.
Boardsworth K, et al. Upper limb robotic rehabilitation following stroke: a systematic review and meta-analysis investigating efficacy and the influence of device features and program parameters. Journal of NeuroEngineering and Rehabilitation. 2025;22:164. The dose-matched analysis included 54 studies and 2,744 participants.
Wang L, Li X, Liu Y, Zhang E, Li C. Effects of robotic-assisted upper extremity therapy for stroke patients in different recovery phases: a systematic review and meta-analysis. Frontiers in Neurology. 2026. Forty-two RCTs involving 1,678 participants were analyzed.
Wang H, Wu X, Li Y, Yu S. Efficacy of Robot-assisted Training on Upper Limb Motor Function After Stroke: A Systematic Review and Network Meta-analysis. Archives of Rehabilitation Research and Clinical Translation. 2025. Thirty-one RCTs involving 1,537 participants were included.
Exoskeleton-Assisted Therapy Enhances Upper Limb Motor Recovery in Early Subacute Stroke: A Multicenter, Single-Blind Randomized Controlled Trial. 2026. The trial evaluated exoskeleton-assisted treatment during early-subacute stroke recovery.