From virtual practice to real-world function the transfer problem in digital neurorehabilitation

Digital technologies have expanded opportunities for neurorehabilitation by enabling intensive, interactive, and adaptable motor practice. However, improvements achieved within virtual environments, rehabilitation games, or technology-based assessments do not necessarily translate into meaningful changes in everyday function. This Mini Review examines the transfer problem in digital neurorehabilitation and discusses key factors that may limit the generalization of training-related gains to real-world performance. We synthesize evidence regarding task specificity, sensory and perception-action differences between digital and physical environments, cognitive and contextual demands, practice variability, and patient-specific constraints that influence transfer. Based on these mechanisms, we propose a transfer-oriented framework consisting of five interconnected stages: skill acquisition, adaptive variability, cognitive-motor integration, contextualized real-world practice, and ecological evaluation. This framework emphasizes that transfer should be deliberately targeted rather than assumed as a consequence of successful digital training. We further discuss how virtual reality, augmented reality, wearable sensing, and hybrid rehabilitation models may support different stages of this process when integrated within adaptive and personalized approaches. Future digital rehabilitation research should move beyond evaluating performance within trained tasks and incorporate measures of functional and ecological transfer, including real-world mobility, activity, and participation. A shift from technology-centered rehabilitation toward transfer-oriented design may enhance the clinical impact of digital interventions and support meaningful improvements in daily life.