Sensor-Based Dataset for Knee Joint Mobility and Rehabilitation Monitoring During Climbing and Walking Activities with Extracted Features
Description
The Sensor-Based Dataset for Knee Joint Mobility and Rehabilitation Monitoring includes four distinct files, each serving a specific role in analyzing knee function during movement. The Walking Data file holds raw sensor readings collected during regular and brisk walking, helping to assess gait characteristics and joint mobility. The Climbing Data captures sensor output from stair-climbing tasks, offering valuable insights into knee performance under increased physical effort.Data is collected from 150 persons for both the activities for the duration of 2 minutes. To improve model training, the Augmentation Data contains artificially varied versions of the original signals created through methods like time shifting and noise injection. Lastly, the Feature Extraction Data provides a structured collection of calculated metrics such as minimum value, maximum value,mean,variance,skewness and kurtosis, which are useful for evaluating movement patterns and supporting further analysis in rehabilitation contexts. Together, these files form a robust foundation for research and development in wearable health monitoring systems.
Files
Steps to reproduce
Development of Smart Knee Belt A wearable knee belt was specifically created for the study, incorporating two MPU6050 sensors placed above and below the knee joint to track movement accurately. Sensor Capabilities Each sensor includes both an accelerometer and a gyroscope, allowing it to record changes in position and rotational motion across three axes. Activity-Based Data Recording Participants wore the belt while performing walking and stair climbing tasks. Each activity was monitored for a duration of 2 minutes. Sampling Frequency and Data Points Motion data was recorded at consistent intervals, resulting in 120 samples per activity per participant. Collection of Participant Details Basic participant information such as age, gender, and BMI was recorded to enable demographic analysis and enhance the relevance of the dataset. Data Augmentation for Diversity Techniques such as noise addition and time shifting were used to generate additional data samples, improving the dataset's variety and supporting robust model training. File Format and Organization All collected and processed data was saved in CSV format, ensuring ease of use and compatibility with various data analysis tools.