International Consensus on Robotic Systems for Stroke Treatment

The first international consensus on the design, testing, and evaluation of robotic systems for stroke treatment has been established, marking a significant step toward standardizing the use of robotics in neurorehabilitation and acute stroke care. The consensus, developed by a global panel of clinicians, engineers, and regulatory experts, provides unified guidelines to ensure safety, efficacy, and reproducibility in robotic interventions for stroke survivors.

Published in Nature Medicine on April 15, 2026, the consensus statement emerged from a two-year collaborative effort led by the World Stroke Organization (WSO) in partnership with the International Society for Physical and Rehabilitation Medicine (ISPRM) and the IEEE Engineering in Medicine and Biology Society. Over 120 experts from 30 countries contributed to the framework, which addresses critical gaps in how robotic technologies are assessed and deployed across diverse healthcare settings.

Stroke remains a leading cause of long-term disability worldwide, affecting more than 12 million people annually. While robotic-assisted therapy has shown promise in improving motor recovery—particularly for upper limb function—variability in device design, testing protocols, and clinical evaluation has hindered widespread adoption and comparability of outcomes across studies and clinics.

Without standardized approaches, it becomes difficult to determine whether a robotic system is truly effective or simply producing inconsistent results due to poor testing methods or mismatched patient selection.

— Dr. Elena Rossi, lead author and neurologist at the University of Milan, speaking at the consensus launch

The consensus outlines three core domains: design principles, preclinical and clinical testing requirements, and real-world evaluation metrics. For design, it recommends that robotic systems be developed with clear biomechanical rationale, adjustable assistance levels, and safety mechanisms such as force limits and collision detection. Devices should also prioritize usability for both patients and therapists, including intuitive interfaces and adaptability to varying levels of motor impairment.

A robot that cannot be safely used by a therapist with minimal training or that causes discomfort due to rigid movement patterns will fail in real-world rehabilitation settings, no matter how advanced its sensors or algorithms.

— Consensus Statement, Section 4.1: Design Principles for Clinical Translation

On testing, the framework mandates a tiered validation process. Preclinical testing must include biomechanical modeling, benchtop durability assessments, and safety simulations using synthetic or animal models before human trials. Clinical testing should progress from feasibility studies in small cohorts to larger randomized controlled trials (RCTs) with clearly defined primary outcomes, such as changes in Fugl-Meyer Assessment scores or Action Research Arm Test performance.

The consensus emphasizes the importance of controlling for confounding variables in trials, including concomitant therapy dosage, time since stroke onset, and baseline impairment severity. It also calls for standardized reporting of adverse events, device usage logs, and therapist interaction times to improve transparency and reproducibility.

We’ve seen too many studies where robot-assisted therapy appears beneficial, but it’s unclear whether the gains came from the robot itself or from the increased therapy time it enabled. This consensus pushes for designs that isolate the robotic contribution.

— Dr. Rajesh Patel, biomedical engineer at Johns Hopkins University and co-chair of the testing subcommittee

For real-world evaluation, the consensus recommends tracking functional outcomes over at least three months post-intervention, assessing not only motor recovery but also quality of life, independence in daily activities, and caregiver burden. It encourages the use of digital health tools—such as wearable sensors and smartphone-based assessments—to capture real-world usage and adherence outside clinical environments.

Recovery doesn’t end when the patient leaves the clinic. A truly useful robotic system should support continued practice at home, and we need ways to measure that meaningfully.

— Consensus Statement, Section 5.3: Community and Long-Term Engagement

The guidelines also address equity and access, urging developers to consider cost-effectiveness, scalability, and adaptability to low-resource settings. The consensus notes that high-cost, complex systems may limit adoption in community hospitals or developing regions, where stroke burden is often highest. It advocates for modular designs and open-source control platforms where feasible to promote broader accessibility.

Innovation in stroke robotics must serve not just well-funded urban centers, but also rural clinics and overburdened public health systems. The best technology is the one that can be used.

— Dr. Amina Diallo, public health specialist at the African Stroke Organization

The consensus does not endorse specific devices or brands but instead provides a benchmark against which new and existing systems can be evaluated. It is intended to inform regulatory agencies, hospital procurement committees, and research funders about what constitutes rigorously developed and validated robotic technology for stroke care.

Organizations such as the U.S. Food and Drug Administration (FDA), the European Medicines Agency (EMA), and Japan’s Pharmaceuticals and Medical Devices Agency (PMDA) have been consulted during the consensus process, and alignment with existing regulatory frameworks for medical devices and software is emphasized. The authors suggest that adherence to these guidelines could streamline future regulatory submissions and health technology assessments.

While the consensus marks a milestone in the field, experts acknowledge that robotic stroke therapy remains an evolving area. Long-term data on durability of functional gains, optimal dosing schedules, and integration with emerging technologies like brain-computer interfaces or non-invasive brain stimulation are still needed. Future updates to the guidelines are planned every three to five years, or as significant technological shifts occur.

As robotic systems become more sophisticated—incorporating artificial intelligence for adaptive assistance, virtual reality for engagement, and soft robotics for safer interaction—the need for clear, evidence-based standards grows. This international consensus offers a foundational step toward ensuring that innovation in stroke rehabilitation translates into real, measurable benefits for patients across the globe.