Implementing Rehabilitation Robotics in Routine Care: A Mapping Review of Service Delivery Models, Implementation Determinants, Strategies, and Real-World Outcomes

Background/Objectives: Rehabilitation robotics is increasingly used in neurorehabilitation, yet evidence on how robotic services are organized, maintained and scaled in routine care remains fragmented. This mapping review examined service-delivery models, implementation determinants, strategies, outcomes, workflows and resource signals for robotic rehabilitation in service-linked clinical care. Methods: Six search sources were queried from inception to 21 February 2026. Eligible reports were studies addressing rehabilitation robotics within clinical pathways and reporting an implementation-relevant element. Data from 55 studies were systematically charted and mapped to CFIR determinants, ERIC strategies and Proctor implementation outcomes. Design-appropriate appraisal was applied once per included report to contextualize methodological confidence. Results: Staffing and supervision were reported in 50 studies (90.9%), workflow and scheduling in 43 (78.2%), safety governance in 32 (58.2%) and training and competency in 29 (52.7%). Feasibility was mapped in 40 studies (72.7%), acceptability in 31 (56.4%) and adoption in 23 (41.8%), whereas sustainability, penetration, technical support, documentation infrastructure and implementation cost were less consistently reported. Two randomized trials were appraised with RoB 2 and four non-randomized comparative studies with ROBINS-I; the remaining 49 reports received design-appropriate methodological or reporting appraisal. Conclusions: Routine-care rehabilitation robotics is best understood as a service configuration rather than a device alone. Transferable implementation requires explicit reporting of workforce, workflow, safety, documentation, technical support, infrastructure, cost, equity and sustainment. Frequencies indicate reporting presence, not practical importance or causal influence. The proposed workflow, logic model and minimum reporting set are evidence-informed, inductively derived proposals requiring prospective validation.