Roessingh Research & Development-MyLeg database for activity prediction (MyPredict)

Roessingh Research &amp; Development-MyLeg database for activity prediction (MyPredict). The general aim of this database is to promote research in data-driven intent recognition strategies and activity prediction strategies in the lower-limb using electromyography and to promote research and development in the area of multi-array sEMG in the lower limb. The database contains three data sets, each containing kinematics and sEMG from able-bodied subjects. In total 55 subjects participated over 85 measurement sessions. Each data set contained a different sEMG measuring protocol containing either traditional bipolar sEMG or multi-array sEMG or a combination of both. In these data sets the subjects transitioned freely from one activity to the next, providing challenging data sets for activity recognition and providing the possibility to investigate human kinematics and sEMG during gait-related activities. This dataset is described in detail in <em>Database of lower limb kinematics and electromyography during gait-related activities in able-bodied subjects</em> (Schulte et al.)<br>MyPredict consists of three datasets, denoted by MP1XX, MP2XX and MP3XX.MyPredict 1: MP101-MP110, 10 able-bodied subjects (sex: 7m, 3f; age: 24±2 years; weight: 77±10 kg; height: 183±9cm), measured onceMyPredict 2: MP201-MP235, 35 able-bodied subjects (sex: 14m, 21f; age: 23±2 years; weight: 73±11 kg; height: 179±9 cm), measured onceMyPredict 3: MP301-MP310, 10 able-bodied subjects (sex: 4m, 6f; age: 24±2 years; weight: 71±9 kg; height: 174±6 cm), measured 4 times<br>These files contain the measurement moment named `Day_X' with X the number of the measurement moment. Inside these measurement moments there are files called `Trial_YY', with YY the trial number, containing the different data types and `MVC' containing the EMG maximum voluntary contractions of each measurement moment. Note that only MyPredict 3 contains multiple measurement moments per subject.<br>The different data types are acceleration (Acc), angular velocity (Gyr), joint angles (Ang), Orientation (Ori) and electromyography (EMG). Inside each file there are trials containing data arrays with the corresponding data. Data arrays are named as follows: Type_Side_Loc. Type is one of the six data types, Loc is the location of the sensor and Side is the side of the location, either Left, Right or empty. For example Ang_Right_Knee contains the 3D joint angles of the knee, Gyr_Pelvis contains the 3D angular velocity of the pelvis IMU and EMG_Left_VL contains the EMG data of the left vastus lateralis. Orientation is the orientation of the pelvis in space, expressed in Euler angles. Separate data types are 'Labels', which contains manual placed activity labels for each timestamp and 'Time' which indicates the timestamps per file. Marker data (Mrk) are stored in a separate group, `Markers' with their own `Time' array, as they have a different sample frequency (100Hz) compared to the other data types (1000Hz).<br>Code supporting this dataset can be found in the github repository: github.com/Rvs94/MyPredict<br><br><br>

Roessingh研究与发展中心的用于活动预测的MyLeg数据库（MyPredict）。本数据库的总体目标是推动基于肌电信号的下肢数据驱动意图识别策略与活动预测策略的研究，并促进下肢多阵列表面肌电图（surface electromyography, sEMG）领域的研究与开发。
该数据库包含三个子数据集，均收录了健康受试者的运动学数据与表面肌电图（sEMG）数据。总计有55名受试者参与了85次测量场次。每个子数据集采用不同的表面肌电图采集方案，涵盖传统双极表面肌电图、多阵列表面肌电图，或二者的组合。在这些数据集中，受试者可自由切换不同活动，为活动识别任务提供了具有挑战性的数据集，同时也为研究步态相关活动中的人体运动学与表面肌电信号特征提供了可能。该数据集的详细信息可参见《健康受试者步态相关活动期间的下肢运动学与肌电图数据库》（Schulte等人）。
MyPredict包含三个子数据集，分别记为MP1XX、MP2XX与MP3XX：
MyPredict 1：对应MP101-MP110，包含10名健康受试者（性别：7男，3女；年龄：24±2岁；体重：77±10 kg；身高：183±9cm），仅完成1次测量。
MyPredict 2：对应MP201-MP235，包含35名健康受试者（性别：14男，21女；年龄：23±2岁；体重：73±11 kg；身高：179±9 cm），仅完成1次测量。
MyPredict 3：对应MP301-MP310，包含10名健康受试者（性别：4男，6女；年龄：24±2岁；体重：71±9 kg；身高：174±6 cm），每名受试者完成4次测量。
相关文件以`Day_X`命名，其中X代表测量场次序号。每个测量场次下包含名为`Trial_YY`的文件（YY为试次编号），存储各类数据，以及名为`MVC`（最大自主收缩，Maximum Voluntary Contraction, MVC）的文件，存储各测量场次的肌电图最大自主收缩值。需注意，仅MyPredict 3包含每名受试者的多次测量场次。
不同的数据类型包括加速度（Acceleration, Acc）、角速度（Angular velocity, Gyr）、关节角度（Joint Angles, Ang）、方位（Orientation, Ori）与肌电图（electromyography, EMG）。每个文件内的试次均包含对应的数据数组，数据数组命名规则为：Type_Side_Loc。其中Type为上述六种数据类型之一，Loc为传感器位置，Side为采集侧别，可选左侧（Left）、右侧（Right）或无侧别。例如，`Ang_Right_Knee`代表膝关节的三维关节角度数据，`Gyr_Pelvis`代表骨盆惯性测量单元（Inertial Measurement Unit, IMU）的三维角速度数据，`EMG_Left_VL`代表左侧股外侧肌的肌电数据。方位（Ori）指骨盆在空间中的方位，以欧拉角表示。
额外的数据类型包括`Labels`，存储各时间戳对应的人工标注活动标签，以及`Time`，存储各文件的时间戳信息。标记数据（Mrk）存储于单独的`Markers`分组中，该分组拥有独立的`Time`数组，原因是其采样频率为100Hz，与其他数据类型的1000Hz采样率不同。
支持该数据集的代码可在GitHub仓库获取：github.com/Rvs94/MyPredict