Stroke is the leading cause of acquired disability in adults in Spain and Europe. Each year, approximately 120,000 people suffer a stroke in the country, and a significant proportion of them will present an upper limb motor deficit that will limit their ability to perform daily activities such as dressing, eating or personal hygiene[1]. Rehabilitation of the affected arm and hand after stroke is one of the greatest clinical challenges in neurorehabilitation, and robotic technology has emerged over the past two decades as a tool with potential to improve the functional outcomes of these patients.
This article reviews the available scientific evidence on robotic upper limb therapy after stroke, the principles that underpin it, the most widely used clinical protocols and the approach we adopt at GNeuro, a robotic neurorehabilitation centre in Ourense.
Upper limb paresis is one of the most frequent and disabling consequences of stroke. Epidemiological data indicate that between 70% and 80% of stroke survivors present arm and hand weakness in the acute phase[2]. Of these, only 12% to 34% achieve complete functional recovery of the upper limb at six months, according to the review by Kwakkel et al.[3]. This means that the majority of people who suffer a stroke with arm involvement will retain some degree of functional limitation long-term.
The functional impact of upper limb paresis is considerable. The hand and arm are involved in virtually all activities of daily living: from feeding and dressing to personal hygiene and written communication. Loss of upper limb function is directly associated with reduced independence, greater need for third-party assistance and a significant reduction in quality of life. Furthermore, «learned non-use» —a phenomenon by which the patient progressively stops attempting to use the affected arm— can worsen the motor deficit if not actively and promptly addressed[4].
This clinical reality justifies the search for therapeutic strategies that allow increasing the intensity of upper limb motor training and maximising the neuroplasticity potential of the central nervous system after injury.
Intensive repetition. Use-dependent neuroplasticity requires a high number of movement repetitions to induce cortical-level changes. Studies in animal models and humans have shown that hundreds, or even thousands, of repetitions are needed to promote brain reorganisation[5]. In a conventional physiotherapy session, the number of repetitions achieved is usually limited (often fewer than 50 functional movements per session). The upper limb rehabilitation robot can facilitate the patient performing between 500 and 1,000 repetitions per session, multiplying the motor training dose.
Controlled and adjustable intensity. Robotic devices allow precise adjustment of the level of assistance the patient receives during movement. From full assistance —where the robot guides the entire joint range— to progressive resistance where the patient must overcome the device's opposition. This gradual adjustment allows adapting difficulty to the patient's residual capacity at each point in their evolution.
Real-time feedback. Robotic systems incorporate sensors that measure forces, positions and movement speeds. This information is presented to the patient visually, auditorily or haptically, providing immediate feedback that promotes motor learning. Many devices integrate video game-type interfaces that increase motivation and treatment adherence.
«Assist-as-needed» assistance. The most advanced control algorithms apply the principle of minimal necessary assistance: the robot only intervenes when it detects the patient cannot complete the movement alone. This strategy promotes active participation of the subject, a key factor for neuroplasticity and motor learning[6].
Task-oriented training. Modern robotic rehabilitation protocols incorporate meaningful functional tasks —such as reaching an object, following a trajectory or simulating daily life movements— rather than isolated joint patterns, thus aligning training with the principles of task-oriented motor learning.
The Cochrane systematic review by Mehrholz et al., updated in 2020, constitutes the most comprehensive evidence synthesis on the subject[1]. This meta-analysis included 45 randomised clinical trials with a total of 1,619 participants. Its main findings were:
«Electromechanical and robot-assisted arm training may improve activities of daily living, arm function, and arm muscle strength after stroke.» — Mehrholz et al., Cochrane Database Syst Rev, 2020[1]
The European Stroke Organisation clinical guidelines, in their 2025 update, include robot-assisted therapy among the recommended interventions for post-stroke upper limb rehabilitation[7]. Specifically, ESO suggests considering robotic therapy as a complement to conventional physiotherapy in patients with moderate to severe arm motor deficit, especially when the aim is to increase training intensity. The recommendation is formulated with a moderate level of evidence, acknowledging that benefits are clearer in terms of motor function than in direct transfer to activities of daily living.
The American Heart Association and American Stroke Association guidelines for adult stroke rehabilitation recognise that robotic therapy can be useful for increasing upper limb training intensity[8]. These guidelines emphasise the importance of treatment intensity and dose as determinant factors in functional outcome, and position robotics as one of the tools that can contribute to achieving the necessary practice volumes.
The RATULS trial (Robot Assisted Training for the Upper Limb after Stroke), published by Rodgers et al. in 2019, was one of the largest multicentre trials on robotic upper limb rehabilitation[9]. With 770 participants, it compared robotic therapy with an intensive conventional upper limb therapy programme and with usual care. Results did not show superiority of robotics over the intensive conventional programme, but did show superiority over usual care. This finding reinforces the idea that the benefit of robotics lies fundamentally in its capacity to increase training intensity, rather than in an intrinsic advantage of the device over the therapist's manual intervention.
The study by Lo et al. (2010), published in the New England Journal of Medicine, demonstrated that robotic arm therapy produced significant improvements in upper limb motor function at 12 months in patients with chronic stroke, compared to usual care[10]. This study was relevant because it showed that neuroplasticity and functional improvement capacity can be maintained even in chronic phases of stroke, provided a sufficient training stimulus is supplied.
Kwakkel et al. (2008) published a meta-analysis in the Journal of NeuroEngineering and Rehabilitation analysing the relationship between robotic training intensity and motor outcomes[3]. Their conclusions indicated a positive dose-response relationship: the greater the number of repetitions and active practice time, the greater the probability of functional improvement. This finding is consistent with the principles of use-dependent neuroplasticity and supports the use of robotics as a tool to increase the therapeutic dose.
Although there is variability between centres and devices, robotic upper limb rehabilitation protocols after stroke share a series of common characteristics based on available evidence:
Programme duration: between 4 and 12 weeks, depending on the stroke phase, functional objectives and patient tolerance. Clinical trials typically use 6 to 8-week programmes.
Frequency: 3 to 5 sessions per week. Evidence suggests greater frequency is associated with better outcomes, provided the patient's recovery capacity between sessions is respected.
Session duration: between 30 and 60 minutes of active work with the robot, in addition to preparation time, device fitting and cool-down.
Number of repetitions per session: between 500 and 1,000 movement repetitions, organised into task blocks with rest intervals.
Progression: protocols follow a gradual progression logic. In early sessions, high robot assistance is used to facilitate movement. As the patient improves, assistance is reduced and active demand increases, progressing from guided to assisted movements, and finally to resisted movements.
Training components: protocols typically include reaching exercises, forearm pronation-supination, wrist flexion-extension and, where the device allows, hand opening and closing. The combination of proximal (shoulder and elbow) and distal (wrist and hand) training appears to produce better functional results than isolated training of a single segment[6].
One of the most consistent findings in neurorehabilitation literature is the positive relationship between training intensity and functional outcomes. Current clinical guidelines recommend that stroke patients receive as much therapy as they can tolerate, at least during the first weeks and months after the event[7][8].
In the specific context of robotic upper limb rehabilitation, evidence suggests:
It is important to emphasise that intensity should be adapted individually. Not all patients tolerate the same training load, and factors such as fatigue, pain, cognitive alterations or medical comorbidities must be taken into account when dosing therapy.
Referral to a robotic upper limb rehabilitation programme may be considered in the following clinical scenarios:
Referral should be made by the rehabilitation physician in coordination with the physiotherapist, following an individualised assessment including evaluation of motor function (Fugl-Meyer Assessment, Action Research Arm Test), cognitive capabilities, exercise tolerance and functional objectives of the patient.
At GNeuro, a robotic neurorehabilitation centre located in Ourense, we have an upper limb rehabilitation robot that allows intensive upper limb training in patients with neurological damage. This device covers the entire upper extremity —from the shoulder to the hand— and offers multiple degrees of freedom to work complex functional movements in a three-dimensional environment.
Our approach is based on the following principles:
Individualised assessment. Before beginning any robotic rehabilitation programme, each patient is comprehensively evaluated by the clinical team. Motor function, sensitivity, muscle tone, cognitive capabilities and functional objectives are assessed. This assessment allows designing a training programme adapted to the specific needs of each person.
Integration within a complete rehabilitation programme. Robotic therapy is not applied in isolation, but as a complement to an individualised programme that may include manual physiotherapy, occupational therapy, speech and language therapy and neuropsychology, according to each patient's needs. Evidence is clear: the best results are obtained when robotics is integrated in a multidisciplinary approach[1][7].
Objective progress monitoring. The upper limb rehabilitation robot automatically records parameters such as strength, speed, movement precision and joint range in each session. This data allows quantifying patient progress objectively, adjusting the programme in real time and sharing precise information with the responsible rehabilitation physician.
Specialisation in neurological damage. GNeuro is a centre specialising exclusively in robotic neurorehabilitation. Our entire clinical team has specific training and experience in managing patients with neurological damage —stroke, traumatic brain injury, spinal cord injury— which allows a level of specialisation that can contribute to optimising treatment outcomes.
It is essential to be transparent about the current limitations of robotic upper limb therapy:
When can robotic arm rehabilitation begin after a stroke?
Robotic upper limb rehabilitation can begin once the patient is clinically stable, usually from the first weeks after the stroke. Clinical guidelines recommend starting neurorehabilitation as soon as possible, ideally within the first 4 to 12 weeks, to take advantage of the window of greatest neuroplasticity. The specific decision should be made by the rehabilitation team based on medical status, exercise tolerance and functional objectives of each person.
Does robotic arm therapy replace conventional physiotherapy?
No. Robotic upper limb therapy does not replace conventional physiotherapy, but complements it. Scientific evidence, including Cochrane reviews and European Stroke Organisation guidelines, indicates that the best results are obtained when robotics is integrated within an individualised rehabilitation programme that includes manual physiotherapy, occupational therapy and, where appropriate, speech and language therapy. The robot allows increasing the intensity and number of repetitions of movements, but the clinical reasoning of the physiotherapist remains fundamental to adapting the treatment.
Which patients benefit most from robotic arm rehabilitation?
Evidence suggests that patients with moderate to severe upper limb motor deficit are those who may benefit most from robot-assisted therapy. Specifically, people who retain some degree of muscle activation but do not have sufficient strength or control to perform functional movements independently. In these cases, the upper limb rehabilitation robot can provide the necessary assistance to complete the movement and train motor patterns that could otherwise not be practised with the required intensity.
How many robotic arm rehabilitation sessions are needed to notice improvement?
There is no fixed number of sessions that guarantees results, as progress depends on multiple factors: stroke severity, time elapsed, motivation and general condition of the patient. However, the protocols used in clinical trials typically include between 20 and 40 sessions distributed over several weeks, with a frequency of 3 to 5 sessions per week. What evidence does show is that a higher number of repetitions and greater practice intensity are associated with better functional outcomes.
If you or a family member has suffered a stroke and wishes to know how robotic upper limb rehabilitation can contribute to their treatment programme, our clinical team can carry out an individualised assessment.
Request an assessment