Intelligent Robot-Aided, Task-Oriented Physiotherapy Method for Upper Limb Rehabilitation: Development and Usability Study.

Single-session trial with no pre- or post-assessment of function; no Fugl-Meyer, ARAT, or any standardised outcome measure was used. Participants were a heterogeneous mix of cardiological, neurological, and orthopaedic conditions, not a targeted stroke or neurological cohort. No randomisation of mode order; every participant did all three modes in the same session, so order effects and fatigue are uncontrolled. Physiotherapist-guided sessions were consistently shorter than robot-assisted ones, which inflates the iEMG and activation-time differences and makes the comparison partly a comparison of duration and postural demand rather than therapeutic effect. No family-wise error correction was applied across the many per-channel, per-exercise comparisons; the authors themselves flag the results as exploratory. The exoskeleton's mechanical design limited some degrees of freedom (open sliding bearings, wrist free but constrained), so ROM in the robot modes may be lower than what the patient could achieve unaided. The authors acknowledge that the higher EMG in robot modes may partly reflect the stabilising demand of the harness rather than a therapeutic advantage.

Funded by the National Center for Research and Development (LIDER XIV, contract LIDER14/0196/2023), 1,793,900 PLN, 2024–2026, for the SmartEx-Home project. No conflicts of interest declared by the authors.

Do not read the significantly higher EMG in the robot modes as proof the robot is a better therapist. The physiotherapist-guided sessions were consistently shorter, the exoskeleton harness forced a stabilising posture that kept the latissimus dorsi and pectoralis major active throughout, and no functional outcome (Fugl-Meyer, ARAT, ADL independence) was measured. The paper shows the robot can drive muscle to work; it does not show the patient recovers better.