Related Experiment Video
Updated: May 31, 2026
07:53
Assessment of Neuromuscular Function Using Percutaneous Electrical Nerve Stimulation
Published on: September 13, 2015
Predict neuromuscular performance in human epidural electrical stimulation: phase 1 trial interim results
Hongda Li1,2, Yunyue Wei2, Yanan Sui2
1National Engineering Research Center of Neuromodulation, School of Aerospace Engineering, Tsinghua University, Beijing, China.
Communications Medicine
|May 28, 2026
Summary
This study developed an AI framework to predict muscle responses to epidural electrical stimulation (EES) for spinal cord injury patients. The AI efficiently identified optimal EES parameters, restoring motor function and reducing clinical trial and error.
Area of Science:
- Neuroscience and Biomedical Engineering
- Computational modeling for neuromodulation therapies
Background:
- Epidural electrical stimulation (EES) shows promise for motor function restoration in paralysis.
- Identifying optimal EES parameters is challenging due to complex stimulation-response relationships.
- A computational framework is needed to predict neuromuscular performance and streamline parameter selection.
Purpose of the Study:
- To develop a computational framework for predicting neuromuscular performance under EES.
- To reduce the extensive in-clinic parameter searches required for EES therapy.
- To optimize stimulation parameters for specific motor objectives in individuals with spinal cord injury.
Main Methods:
- Implanted 32-contact epidural interfaces in two individuals with motor-complete spinal cord injury.
- Integrated finite element simulations and axonal recruitment modeling with machine learning for predictive mapping.
- Applied a dimensionality-reduction Bayesian optimization algorithm to identify stimulation parameters, validated clinically.
Main Results:
- Purpose-designed epidural interfaces enabled lower-limb motor function recovery in two participants.
- The predictive model showed high accuracy, with a mean squared error of 0.0096 compared to experimental data.
- AI-guided parameter recommendations (180 configurations) outperformed historical data (1,602 configurations) and conventional settings across four functional objectives.
Conclusions:
- An AI-aided computational framework reliably evaluates and recommends effective EES parameters.
- This approach bridges anatomical modeling with functional outcomes for optimizing neuromodulation.
- It offers a pathway for personalized treatment strategies in spinal cord injury rehabilitation.
