Lower Limb Joints Torques Continuous Estimate Model Based on Muscle Synergy for Patients With Motor Dysfunction

Area of Science:

• Biomedical Engineering
• Robotics
• Neuroscience

Background:

• Lower limb joint torque estimation is vital for effective rehabilitation exoskeleton control.
• Surface electromyography (sEMG) offers insights into human motion for assistive technologies.

Purpose of the Study:

• To develop a neural control-level model linking myoelectric signals to lower limb joint torques.
• To enhance the accuracy and long-term adaptability of torque estimation for rehabilitation robots.

Main Methods:

• Established a muscle bioelectrical activation model from sEMG signals.
• Developed a motor nerve activation model incorporating muscle synergy.
• Integrated deep learning with self-attention and adversarial transfer learning for continuous torque estimation.

Main Results:

• The model accurately estimated hip and knee joint torques in eight patients with lower limb motor dysfunction.
• Achieved high performance over long-term use with decision coefficients of 0.92 ± 0.06 (hip) and 0.95 ± 0.03 (knee).

Conclusions:

• The proposed model provides a reliable method for estimating lower limb joint torques using sEMG.
• This research paves the way for advanced, on-demand assisted control in exoskeleton-based rehabilitation.