Related Experiment Videos
Predicting joint moments and angles from EMG signals
1Department of IMSE, College of Engineering, Iowa State University, Ames 50011, USA.
Biomedical Sciences Instrumentation
|January 1, 1997
Summary
This study developed a neural network to predict wheelchair propulsion movement from electromyographic (EMG) signals. The model accurately estimated joint dynamics, showing potential for understanding wheelchair user biomechanics.
Area of Science:
- Biomechanics
- Biomedical Engineering
- Neuroscience
Background:
- Electromyographic (EMG) signals reflect muscle activity but extracting kinematic information is challenging.
- Previous attempts to correlate EMG signals with kinematic outcomes have yielded limited success.
Purpose of the Study:
- To develop a neural network model correlating EMG signals with kinematic features of the wrist, elbow, and shoulder joints during wheelchair propulsion.
- To predict joint dynamics using previously unseen EMG data.
Main Methods:
- Collected EMG signals from four muscle groups in five able-bodied subjects propelling a wheelchair.
- Videotaped the experiment to capture joint kinematics and dynamics.
- Utilized a back-propagation neural network trained on the collected EMG and kinematic data.
Main Results:
- The neural network model accurately predicted joint dynamics from EMG signals.
- Predicted joint moments were within 7% of observed values.
- Demonstrated strong correlation between EMG signals and joint kinematics.
Conclusions:
- Neural networks show significant potential for predicting wheelchair user movement patterns from EMG data.
- The developed model is effective even with a limited number of subjects.
- This approach can advance the understanding of wheelchair biomechanics and user interaction.