Related Experiment Video
Updated: Jul 15, 2025
11:18
Closed-loop Neuro-robotic Experiments to Test Computational Properties of Neuronal Networks
Published on: March 2, 2015
10.3K
The bionic neural network for external simulation of human locomotor system
Arxiv
|September 25, 2023
Summary
This study introduces a physics-informed deep learning approach for musculoskeletal (MSK) modeling. The method accurately predicts joint motion and muscle forces by integrating MSK models into neural networks, overcoming computational challenges.
Area of Science:
- Biomechanics
- Computational modeling
- Deep learning
Background:
- Musculoskeletal (MSK) modeling estimates movement quality using muscle forces and joint kinematics.
- Model-based MSK models face computational challenges and muscle recruitment issues in complex scenarios.
- Data-driven methods offer flexibility but require extensive labeled training data.
Approach:
- Proposes a physics-informed deep learning method integrating MSK models into neural networks.
- Embeds the MSK model as an ordinary differential equation (ODE) loss function for parameter identification.
- Automatically estimates physiological parameters of muscle activation and contraction dynamics during training.
Key Points:
- The method accurately predicts joint motion and muscle forces.
- Effectively identifies subject-specific MSK physiological parameters.
- Validated on benchmark and self-collected datasets from healthy subjects.
Conclusions:
- The proposed deep learning method offers an effective solution for accurate MSK modeling.
- Physics-informed neural networks can overcome limitations of traditional MSK models.
- Enables precise prediction of muscle forces and joint kinematics for movement analysis.
