University of Twente Student Theses
As of Friday, 8 August 2025, the current Student Theses repository is no longer available for thesis uploads. A new Student Theses repository will be available starting Friday, 15 August 2025.
Estimation of Knee Joint Kinetics using IMUs and Physics-Informed Machine Learning during Walking and Single-Leg Hop tests in ACLR Rehabilitation
Hofste, Gijs (2025) Estimation of Knee Joint Kinetics using IMUs and Physics-Informed Machine Learning during Walking and Single-Leg Hop tests in ACLR Rehabilitation.
|
PDF
22MB |
| Abstract: | Evaluation of knee joint loading is essential for monitoring recovery and supporting return-to-sport (RTS) decisions following anterior cruciate ligament reconstruction (ACLR). This study investigates the estimation of three-dimensional knee joint kinetics from inertial measurement unit (IMU) data using a physics-informed neural network (PINN) during walking and single-leg hop tests. A data processing pipeline was developed to align and generalize inertial and optical motion capture data. Segment kinematics derived from IMUs were used as input to estimate knee joint kinetics, while enabling three-dimensional inverse dynamics through rigid-body modeling. An adaptable neural network framework was designed to incorporate physical constraints via a loss function based on three-dimensional segment-based inverse dynamics. The PINN was compared to a baseline data-driven model to evaluate prediction performance. While the physical constraints supported biomechanical interpretable estimations, the PINN did not outperform the baseline model. Both models showed reduced accuracy for kinetic components with lower magnitudes, particularly in the mediolateral direction. Moreover, no consistent improvements were found under limited data conditions or varied physical loss weighting. These findings suggest that physical constraints may not enhance performance when data quantity or consistency is insufficient. Despite these limitations, this study provides a methodological basis for future research on IMU-based kinetic estimation in ACLR rehabilitation. Further development is needed to improve model accuracy and robustness. With sufficient performance, such models could enable subject-specific assessments of limb asymmetry and support RTS decision-making, ultimately reducing the risk of reinjury. |
| Item Type: | Essay (Master) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 50 technical science in general |
| Programme: | Biomedical Engineering MSc (66226) |
| Link to this item: | https://purl.utwente.nl/essays/107318 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page