Predicting human gait kinematics and kinetics from a single inertial measurement unit using deep learning and synthetic datasets: a blinded assessment study
File(s)
Published version
Author(s)
Type
Journal Article
Abstract
A single inertial measurement unit (IMU) can be used with neural networks (NNs) to predict joint kinematics and kinetics. Recent studies, however, often employed virtual IMU data computed from marker-based systems to train NN models, without providing realistic performance assessments on real IMU data. To address this, we present a methodology for developing and blindly evaluating NNs predicting lower-limb joint angles and moments from a sacrum-worn IMU.
NN architectures were trained on IMU data simulated from a public marker-based dataset (49 healthy adults), augmented using conditional generative adversarial networks to enhance variability. The developed NNs were blindly tested against a different dataset (seven healthy adults) of real IMU and synchronous marker-based data collected ad hoc after the NN development. The two datasets were collected in different labs using different protocols. These NNs were subsequently fine-tuned (retrained) with this dataset and re-evaluated on another real IMU data (three healthy adults) collected after fine-tuning.
The NNs achieved strong predictive performance on virtual IMU data (average root mean squared error (RMSE) of 2.6±1.3° and 0.10±0.05 Nm/kg for joint angles and moments, respectively). However, performance degraded when applied to real IMU data: average RMSE of 4.5±2.0° for joint angles and 0.21±0.14 Nm/kg for moments. Fine-tuning with real IMU data improved model accuracy, recovering RMSEs to 2.6±0.8° and 0.19±0.11 Nm/kg for joint angles and moments, respectively.
Overall, our performance metrics were within the reported ranges for systems employing multiple IMU sensors. This work highlights the importance of blinded assessment and fine-tuning for practical biomechanical applications.
NN architectures were trained on IMU data simulated from a public marker-based dataset (49 healthy adults), augmented using conditional generative adversarial networks to enhance variability. The developed NNs were blindly tested against a different dataset (seven healthy adults) of real IMU and synchronous marker-based data collected ad hoc after the NN development. The two datasets were collected in different labs using different protocols. These NNs were subsequently fine-tuned (retrained) with this dataset and re-evaluated on another real IMU data (three healthy adults) collected after fine-tuning.
The NNs achieved strong predictive performance on virtual IMU data (average root mean squared error (RMSE) of 2.6±1.3° and 0.10±0.05 Nm/kg for joint angles and moments, respectively). However, performance degraded when applied to real IMU data: average RMSE of 4.5±2.0° for joint angles and 0.21±0.14 Nm/kg for moments. Fine-tuning with real IMU data improved model accuracy, recovering RMSEs to 2.6±0.8° and 0.19±0.11 Nm/kg for joint angles and moments, respectively.
Overall, our performance metrics were within the reported ranges for systems employing multiple IMU sensors. This work highlights the importance of blinded assessment and fine-tuning for practical biomechanical applications.
Date Issued
2026-03-01
Date Acceptance
2026-01-05
Citation
Journal of Biomechanics, 2026, 198
ISSN
0021-9290
Publisher
Elsevier
Journal / Book Title
Journal of Biomechanics
Volume
198
Copyright Statement
© 2026 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
License URL
Identifier
https://www.ncbi.nlm.nih.gov/pubmed/41687508
PII: S0021-9290(26)00003-5
Subjects
Blinded assessment
Gait
Inertial measurement unit
Synthetic mocap dataset
Humans
Biomechanical Phenomena
Deep Learning
Gait
Adult
Male
Female
Kinetics
Neural Networks, Computer
Young Adult
Publication Status
Published
Coverage Spatial
United States
Article Number
113149
Date Publish Online
2026-01-06