Related Experiment Video
Updated: Feb 28, 2026
08:24
Sit-to-stand-and-walk from 120% Knee Height: A Novel Approach to Assess Dynamic Postural Control Independent of Lead-limb
Published on: August 30, 2016
10.8K
Long Short-Term Memory Network for Contralateral Knee Angle Estimation During Level-Ground Walking: A Feasibility
Ala'a Al-Rashdan1, Hala Amari1, Yahia Al-Smadi2,3
1Biomedical Engineering Department, Faculty of Engineering, Jordan University of Science and Technology, Irbid 22110, Jordan.
Micromachines
|February 27, 2026
Summary
Researchers developed a long short-term memory (LSTM) network to predict knee joint angles using inertial measurement units (IMUs). This technique shows promise for improving prosthetic knee control in transfemoral amputees.
Area of Science:
- Bionics and Biomedical Engineering
- Rehabilitation Engineering
- Artificial Intelligence in Healthcare
Background:
- Rising incidence of lower limb amputations due to war, accidents, and vascular diseases, with transfemoral amputations comprising 39% of cases.
- Critical need for advanced prosthetic knee joints to enhance functional mobility and daily living activities for amputees.
- Accurate prediction of knee joint angle is essential for effective transfemoral prosthesis control during gait.
Purpose of the Study:
- To develop and evaluate a novel technique for estimating knee joint angles using Long Short-Term Memory (LSTM) networks.
- To utilize kinematic data from Inertial Measurement Units (IMUs) for real-time knee angle prediction.
- To investigate the feasibility of LSTM networks in mimicking natural knee biomechanics for prosthetic applications.
Main Methods:
- Training and testing an LSTM network with kinematic data from twenty able-bodied subjects.
- Utilizing a custom sensory gadget equipped with four IMUs to collect data.
- Estimating the contralateral knee angle as a proxy for intended prosthetic knee movement.
Main Results:
- The LSTM models demonstrated high accuracy in estimating contralateral knee joint angles, with a real-time Root Mean Square Error (RMSE) range of 2.48-2.78°.
- A high correlation coefficient range of 0.9937-0.9991 was achieved, indicating robust performance.
- The study confirmed the real-time performance and robustness of LSTM networks for knee angle estimation in able-bodied individuals.
Conclusions:
- LSTM networks are effective for estimating contralateral knee joint angles, showing potential for prosthetic knee control.
- The developed technique demonstrates feasibility and robustness, paving the way for future research in bionic limb development.
- Further clinical validation with amputee participants is necessary to generalize findings and optimize prosthesis control.
