Gait_Posture_dataset

Description

This dataset contains extracted repetition-level biomechanical feature tables and accompanying Python analysis scripts for lower-limb movement tasks measured using three systems: Vicon, Theia, and Femto. The dataset includes feature tables for hip abduction, lunge, and squat tasks from 22 participants across two testing visits. Each row represents one detected movement repetition for a specific participant, visit, set, and measurement system. The extracted variables include joint angles, hip abduction angle, lateral and vertical displacement, and vertical center-of-mass displacement, depending on the task. Angle variables are reported in degrees, and displacement variables are reported in millimeters. The accompanying Python scripts provide the processing and analysis workflow used to generate and summarize the feature data. The scripts include batch feature extraction with a fourth-order zero-phase Butterworth low-pass filter, waveform processing, descriptive statistics, system agreement analysis, Bland-Altman analysis, and reliability analyses. Vicon is treated as the reference system in the reliability and validity analyses. Documentation files are included to describe the dataset structure, variable definitions, units, and analysis code workflow.

Steps to reproduce

1. Download all CSV feature tables, documentation files, and Python scripts. 2. Review README.md, README_Feature_Data.md, Data_Dictionary_Feature_Tables.md, and Analysis_code/README_Mendeley_Data.md to understand the data structure, variable definitions, and analysis workflow. 3. Install the required Python packages: pip install numpy pandas matplotlib seaborn scipy statsmodels 4. Open the Python scripts in the Analysis_code folder and update the local file paths near the top of each script, including FILE_PATH, SAVE_DIR, BASE_DIR, RAW_DATA_PATH, and OUTPUT_PATH, as needed. 5. Use the extracted feature tables directly for statistical analysis: - Hip_Abduction_Master_Database_Detailed_Features.csv - Lunge_Master_Database_Detailed_Features.csv - Squat_Master_Database_Detailed_Features.csv 6. To reproduce the manuscript-style reliability and validity outputs, run the relevant analysis scripts in Analysis_code. Vicon is treated as the reference system. The scripts generate descriptive statistics, agreement metrics, Bland-Altman figures, ICC-based reliability tables, and supplementary outputs. 7. To reproduce the batch feature extraction workflow from raw motion trial CSV files, run the corresponding batch-processing scripts after placing the raw Vicon/Theia and Femto CSV files in the expected subject/visit/set folder structure described in the documentation.

Institutions

• Auckland University of Technology

  Auckland, Auckland

Categories

Licence