Posture-invariant Myoelectric Control with Self-calibrating Random Forests
Author Information
Author(s): Jiang Xinyu, Ma Chenfei, Nazarpour Kianoush
Primary Institution: School of Informatics, The University of Edinburgh, Edinburgh, United Kingdom
Hypothesis
Can a self-calibrating random forest model improve myoelectric control across varying arm positions?
Conclusion
The study demonstrates that a self-calibrating random forest model can effectively adapt to different arm positions, improving myoelectric control performance.
Supporting Evidence
- The random forest model showed high generalisability to varying arm positions.
- Self-calibration improved model performance during real-time testing.
- Participants' muscle activity varied significantly across different arm positions.
Takeaway
This study shows that a smart computer model can learn to control prosthetic devices better, even when the user's arm is in different positions.
Methodology
The study used a self-calibrating random forest model pre-trained on data from multiple users and tested it on new users to evaluate its performance across different arm positions.
Potential Biases
Potential bias may arise from the limited diversity of the participant sample.
Limitations
The model was validated only on limb-intact subjects, and further studies are needed to assess its performance on amputees.
Participant Demographics
Participants included 66 individuals for pre-training (22-43 years old, 12 males, 8 females) and 20 individuals for real-time evaluation (19-31 years old, 10 males, 10 females).
Statistical Information
P-Value
p<0.05
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website