Related Experiment Video
Updated: Aug 6, 2026
06:45
Automated Joint Space Detection Improves Bone Segmentation Accuracy
Published on: November 28, 2025
Muscle synergy-driven deep learning for predicting multiphase knee joint loading in ACL-reconstructed patients
Yi Yuan1, Tianxiao Chen2, Datao Xu2
1Research Academy of Medicine Combining Sports, Ningbo No. 2 Hospital, Ningbo 315010, China; Orthopedic Center, Ningbo No. 2 Hospital, Ningbo 315010, China.
Computer Methods and Programs in Biomedicine
|July 18, 2026
Summary
This study shows a new AI model can accurately estimate knee contact forces after ACL reconstruction using muscle synergies. This method simplifies complex simulations for better joint loading monitoring during rehabilitation.
Area of Science:
- Biomechanics
- Computational modeling
- Rehabilitation science
Background:
- Accurate knee contact force (KCF) estimation is crucial for monitoring joint loading during anterior cruciate ligament reconstruction (ACL-R) rehabilitation.
- Conventional musculoskeletal simulations (e.g., OpenSim) are effective but computationally intensive and complex.
- There is a need for efficient methods to estimate KCF during ACL-R recovery.
Purpose of the Study:
- To evaluate the feasibility of using a neural network framework (SSA-CNN-xLSTM) integrating muscle synergy-derived neuromuscular representations.
- To reproduce OpenSim simulation-derived KCF across different stages of ACL-R rehabilitation.
- To assess the accuracy and applicability of surrogate modeling for KCF estimation.
Main Methods:
- Thirty-three individuals undergoing unilateral ACL-R were assessed at 3, 6, and 12 months postoperatively.
- Surface electromyography and motion capture data were collected during level walking.
- Muscle synergies were extracted using non-negative matrix factorization (NNMF) and used to train SSA-CNN-xLSTM models to predict KCF.
Main Results:
- Both muscle activation and muscle synergy inputs accurately reproduced OpenSim-derived KCF (R² > 0.90) across all rehabilitation stages.
- Muscle synergy-based models showed comparable or superior predictive performance to muscle activation-based models.
- Low-dimensional neuromuscular representations effectively supported surrogate modeling of KCF throughout recovery.
Conclusions:
- The SSA-CNN-xLSTM framework successfully reproduces OpenSim-derived KCF using non-invasive neuromuscular measurements during ACL-R rehabilitation.
- The longitudinal design confirms the framework's applicability despite changing neuromuscular conditions.
- Muscle synergy-informed surrogate modeling offers a promising complementary approach for biomechanical assessment in ACL-R recovery.
