IntelliRehabDS: a physical rehabilitation movements dataset This dataset contains 3D skeleton joints positions of 15 patients and 14 healthy persons while performing a few simple exercises. The data was recorded using a Kinect Sensor. Each exercise has been annotated with the movement type along with a label showing if the movement was executed correctly or not. The annotations were carried out independently by two annotators. The dataset is presented as a collection of .csv files, where each file represents one exercise performed by one person. The nomenclature of the files is the following: SubjectID_DateID_GestureLabel_RepetitionNumber_CorrectLabel_Position.txt SubjectID: id uniquely identifying the person performing the exercise DateID: id identifying the session in which the person was recorded GestureLabel: Label identifying the gesture; possible values are from 0 to 8 RepetitionNumber: Each gesture was repeated several times and this shows the repetition number CorrectLabel: A value of 1 represents a gesture labeled as being correctly executed, while a value of 2 is for a gesture labeled as incorrect Position: Some of the persons performed the gestures sitting on a chair or wheelchair, while others standing 2022-01-06: Update: Patient with id 217 was excluded from the analysis in [1] because they had to stop the exercises mid recording. We have left the partial data recording in the published dataset. The peer-reviewed data descriptor for this dataset has now been published in MDPI Data and can be accessed here: https://www.mdpi.com/2306-5729/6/5/46. Please cite this when using the dataset [1] Miron, A., Sadawi, N., Ismail, W., Hussain, H. and Grosan, C., 2021. IntelliRehabDS (IRDS)—A Dataset of Physical Rehabilitation Movements. Data, 6(5), p.46.

GitHub repository for basic exploratory data analysis: https://github.com/alina-miron/intellirehabds