Effectiveness of Robotic Devices for Medical Rehabilitation: An Umbrella Review.

Background/Objectives: Clinical trials have investigated the efficacy of rehabilitation robotics for various pathological conditions, but the overall impact on rehabilitation practice remains unclear. We comprehensively examined and analyzed systematic reviews (SRs) of randomized controlled trials (RCTs) investigating rehabilitative interventions with robotic devices. Methods: Four databases were searched using term combinations of keywords related to robotic devices, rehabilitation, and SRs. The SR meta-analyses were categorized into "convincing", "highly suggestive", "suggestive", "weak", or "non-significant" depending on evidence strength and validity. Results: Overall, 62 SRs of 341 RCTs involving 14,522 participants were identified. Stroke was most frequently reported (40 SRs), followed by spinal cord injury (eight SRs), multiple sclerosis (four SRs), cerebral palsy (four SRs), Parkinson's disease (three SRs), and neurological disease (any disease causing limited upper- and lower-limb functioning; three SRs). Furthermore, 38, 21, and 3 SRs focused on lower-limb devices, upper-limb devices, and both upper- and lower-limb devices, respectively. Quantitative synthesis of robotic intervention effects was performed by 51 of 62 SRs. Robot-assisted training was effective for various outcome measures per disease. Meta-analyses offering suggestive evidence were limited to studies on stroke. Upper-limb devices were effective for motor control and activities of daily living, and lower-limb devices for walking independence in stroke. Conclusions: Robotic devices are useful for improving impairments and disabilities in several diseases. Further high-quality SRs including RCTs with large sample sizes and meta-analyses of these RCTs, particularly on non-stroke-related diseases, are required. Further research should also ascertain which type of robotic device is the most effective for improving each specific impairment or disability.