Detection of gait variations by using artificial neural networks
Cem Guzelbulut1, Satoshi Shimono2, Kazuo Yonekura1
1Department of Systems Innovation, School of Engineering, The University of Tokyo, Tokyo, Japan.
Biomedical Engineering Letters
|October 14, 2022
Summary
This study models normal walking variations using an artificial neural network, considering personal factors like age and speed. The model accurately predicts gait parameters, aiding in identifying abnormalities and designing assistive devices.
Area of Science:
- Biomechanics
- Human Movement Analysis
- Artificial Intelligence in Healthcare
Background:
- Gait variations are influenced by individual characteristics and abnormalities.
- Understanding normal gait variations is crucial for medical diagnosis and assistive device design.
- Personal parameters like age, sex, height, weight, and walking speed significantly affect gait.
Purpose of the Study:
- To model normal gait variations based on personal parameters.
- To establish relationships between personal parameters and gait parameters.
- To develop a predictive model for normal walking.
Main Methods:
- Utilized a large dataset of walking trials.
- Developed an artificial neural network (ANN)-based gait characterization model.
- Simulated normal walking by incorporating personal parameters into the ANN model.
Main Results:
- The ANN model successfully simulated normal gait parameters influenced by personal factors.
- Predicted gait parameter behavior showed similarity with existing literature.
- Analyzed differences between experimental data and model predictions to identify excessive deviations.
Conclusions:
- The ANN-based gait characterization model effectively represents normal gait parameter behavior.
- The model aids in understanding individual gait variations.
- This approach can assist in identifying gait abnormalities and informing the design of orthotic and prosthetic products.
