Hybrid Systems for Post-Stroke Upper-Limb Rehabilitation: Current Trends and Future Perspectives

Hybrid systems that combine upper-limb robotics with Functional/Neuromuscular electrical stimulation (FES/NMES) aim to integrate mechanical assistance and neuromuscular activation to enhance post-stroke motor learning. A recent review on combined robot–stimulation interventions for the post-stroke upper limb reported promising outcomes, but highlighted marked heterogeneity in platforms, protocols, and outcome measures. In parallel, the Artificial Intelligence (AI)-driven rehabilitation literature identifies, as structural bottlenecks, the shortage of therapists, the difficulty of continuous monitoring, and the subjectivity of assessment, thereby motivating automated systems and data-driven evaluation. This paper proposes an AI-aware perspective on hybrid robot–FES/NMES systems for post-stroke upper-limb rehabilitation, focusing on: (i) a taxonomy of architectures and integration modalities (simultaneous vs. sequential), (ii) key control/decision gaps (triggering robustness, dose personalization, fatigue/comfort management, and clinical transferability), (iii) opportunities for integrating AI at critical points of the loop (intention perception, state estimation, stimulation adaptation, and planning/assessment). The study includes Randomized Controlled Trials (RCTs) and trial protocols that formalize combined approaches (e.g., Electromyography (EMG)-triggered co-actuation, block-wise combinations, and user-centered/wearable solutions), as well as recent contributions introducing Machine Learning (ML) models for fatigue or discomfort and EMG-based classifiers to enable more natural control.