Search for a command to run...
The dataset comprises synchronized multimodal recordings of hand kinematics, muscle activity, and inertial signals acquired from both healthy participants and individuals with transradial amputation. A total of nine movement classes were defined to capture both isolated and coordinated finger actions. These include cyclic flexion and extension of each individual finger, combined movements of the index and middle fingers, coordinated motion of all fingers excluding the thumb, and two functional grasping tasks: a three-finger cup grip and a full-hand (five-finger) grip.All data were collected using the Delsys Trigno wireless sensing system and a Qualisys optical motion capture system. Surface electromyography (sEMG) signals were acquired from 14 channels using Delsys Trigno sensors at approximately 1259 Hz. Inertial measurements, including accelerometer and gyroscope data, were recorded from the same sensor units at 148 Hz. Hand kinematics were captured using the Qualisys motion capture system at 300 Hz, providing continuous three-dimensional marker trajectories and derived finger joint representations.All data streams were fully time-synchronized and recorded at their respective sampling frequencies. The dataset is organized into three primary data formats. The first format consists of JavaScript Object Notation (JSON) files containing the raw synchronized data, including 14-channel EMG signals, inertial measurements, and kinematic recordings. The second format includes MATLAB (.mat) files containing the same data, where finger orientations are represented as joint angles in degrees rather than quaternions. The third format comprises comma-separated values (CSV) files, which provide a simplified representation of the dataset by including only the metacarpophalangeal (MCP) joint angles for each finger.This structured organization enables flexible use of the dataset for a wide range of applications, including signal processing, machine learning, and biomechanical analysis, while supporting both raw data access and simplified feature-level representations.