A wearable sensor-based kinematic dataset collected under standardized rehabilitation tasks from 120 post-stroke...

Stroke frequently results in long-term motor impairments, making effective rehabilitation essential for functional recovery. However, the development of intelligent rehabilitation systems is hindered by the lack of large-scale kinematic data from stroke patients. Here, we present REHAB, a wearable sensor-based kinematic dataset collected from 120 post-stroke patients during a two-week rehabilitation program. The dataset comprises signals recorded from 27 standardized assessment movements and 16 rehabilitation training movements, providing comprehensive limb kinematics together with corresponding task annotations and clinical labels. Detailed descriptions of the data acquisition protocol, sensor configuration, and data organization are provided to ensure reproducibility and facilitate reuse. In addition, systematic quality-control procedures, including clinician-guided acquisition, sensor calibration, signal inspection, and preprocessing standardization, were implemented throughout the data collection pipeline to ensure data quality and consistency. REHAB provides a comprehensive and clinically relevant kinematic resource for stroke rehabilitation research and may support future studies in rehabilitation assessment, movement analysis, wearable sensing, and data-driven intelligent rehabilitation systems.