Estimation of lower limb torque: a novel hybrid method based on continuous wavelet transform and deep learning approach

Biomechanical analysis of the human lower limbs plays a critical role in movement assessment, injury prevention, and rehabilitation guidance. Traditional gait analysis techniques, such as optical motion capture systems and biomechanical force platforms, are limited by high costs, operational complex...

Full description

Saved in:
Bibliographic Details
Main Authors: Shu Xu, Tao Wang, Zenghui Ding, Yu Wang, Tongsheng Wan, Dezhang Xu, Xianjun Yang, Ting Sun, Meng Li
Format: Article
Language:English
Published: PeerJ Inc. 2025-05-01
Series:PeerJ Computer Science
Subjects:
Online Access:https://peerj.com/articles/cs-2888.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850242700356354048
author Shu Xu
Tao Wang
Zenghui Ding
Yu Wang
Tongsheng Wan
Dezhang Xu
Xianjun Yang
Ting Sun
Meng Li
author_facet Shu Xu
Tao Wang
Zenghui Ding
Yu Wang
Tongsheng Wan
Dezhang Xu
Xianjun Yang
Ting Sun
Meng Li
author_sort Shu Xu
collection DOAJ
description Biomechanical analysis of the human lower limbs plays a critical role in movement assessment, injury prevention, and rehabilitation guidance. Traditional gait analysis techniques, such as optical motion capture systems and biomechanical force platforms, are limited by high costs, operational complexity, and restricted applicability. In view of this, this study proposes a cost-effective and user-friendly approach that integrates inertial measurement units (IMUs) with a novel deep learning framework for real-time lower limb joint torque estimation. The proposed method combines time-frequency domain analysis through continuous wavelet transform (CWT) with a hybrid architecture comprising multi-head self-attention (MHSA), bidirectional long short-term memory (Bi-LSTM), and a one-dimensional convolutional residual network (1D Conv ResNet). This integration enhances feature extraction, noise suppression, and temporal dependency modeling, particularly for non-stationary and nonlinear signals in dynamic environments. Experimental validation on public datasets demonstrates high accuracy, with a root mean square error (RMSE) of 0.16 N·m/kg, Coefficient of Determination (R2) of 0.91, and Pearson correlation coefficient of 0.95. Furthermore, the framework outperforms existing models in computational efficiency and real-time applicability, achieving a single-cycle inference time of 152.6 ms, suitable for portable biomechanical monitoring systems.
format Article
id doaj-art-3ae14fcb18c24fafbb47e52c9b43eb96
institution OA Journals
issn 2376-5992
language English
publishDate 2025-05-01
publisher PeerJ Inc.
record_format Article
series PeerJ Computer Science
spelling doaj-art-3ae14fcb18c24fafbb47e52c9b43eb962025-08-20T02:00:13ZengPeerJ Inc.PeerJ Computer Science2376-59922025-05-0111e288810.7717/peerj-cs.2888Estimation of lower limb torque: a novel hybrid method based on continuous wavelet transform and deep learning approachShu Xu0Tao Wang1Zenghui Ding2Yu Wang3Tongsheng Wan4Dezhang Xu5Xianjun Yang6Ting Sun7Meng Li8Science Island Branch, Graduate School of USTC, University of Science and Technology of China, Hefei, Anhui, ChinaInstitute of Intelligent Machines, Chinese Academy of Sciences, Hefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei, Anhui, ChinaInstitute of Intelligent Machines, Chinese Academy of Sciences, Hefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei, Anhui, ChinaScience Island Branch, Graduate School of USTC, University of Science and Technology of China, Hefei, Anhui, ChinaScience Island Branch, Graduate School of USTC, University of Science and Technology of China, Hefei, Anhui, ChinaAnhui Key Laboratory of Advanced Numerical Control & Servo Technology (Cultivating Base), Wuhu, Anhui, ChinaInstitute of Intelligent Machines, Chinese Academy of Sciences, Hefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei, Anhui, ChinaScience Island Branch, Graduate School of USTC, University of Science and Technology of China, Hefei, Anhui, ChinaThe Second Affiliated Hospital, Anhui University of Chinese Medicine, Hefei, Anhui, ChinaBiomechanical analysis of the human lower limbs plays a critical role in movement assessment, injury prevention, and rehabilitation guidance. Traditional gait analysis techniques, such as optical motion capture systems and biomechanical force platforms, are limited by high costs, operational complexity, and restricted applicability. In view of this, this study proposes a cost-effective and user-friendly approach that integrates inertial measurement units (IMUs) with a novel deep learning framework for real-time lower limb joint torque estimation. The proposed method combines time-frequency domain analysis through continuous wavelet transform (CWT) with a hybrid architecture comprising multi-head self-attention (MHSA), bidirectional long short-term memory (Bi-LSTM), and a one-dimensional convolutional residual network (1D Conv ResNet). This integration enhances feature extraction, noise suppression, and temporal dependency modeling, particularly for non-stationary and nonlinear signals in dynamic environments. Experimental validation on public datasets demonstrates high accuracy, with a root mean square error (RMSE) of 0.16 N·m/kg, Coefficient of Determination (R2) of 0.91, and Pearson correlation coefficient of 0.95. Furthermore, the framework outperforms existing models in computational efficiency and real-time applicability, achieving a single-cycle inference time of 152.6 ms, suitable for portable biomechanical monitoring systems.https://peerj.com/articles/cs-2888.pdfInertial measurement units (IMU)Lower limb joint torqueDeep learning approachBi-LSTMContinuous wavelet transform (CWT)
spellingShingle Shu Xu
Tao Wang
Zenghui Ding
Yu Wang
Tongsheng Wan
Dezhang Xu
Xianjun Yang
Ting Sun
Meng Li
Estimation of lower limb torque: a novel hybrid method based on continuous wavelet transform and deep learning approach
PeerJ Computer Science
Inertial measurement units (IMU)
Lower limb joint torque
Deep learning approach
Bi-LSTM
Continuous wavelet transform (CWT)
title Estimation of lower limb torque: a novel hybrid method based on continuous wavelet transform and deep learning approach
title_full Estimation of lower limb torque: a novel hybrid method based on continuous wavelet transform and deep learning approach
title_fullStr Estimation of lower limb torque: a novel hybrid method based on continuous wavelet transform and deep learning approach
title_full_unstemmed Estimation of lower limb torque: a novel hybrid method based on continuous wavelet transform and deep learning approach
title_short Estimation of lower limb torque: a novel hybrid method based on continuous wavelet transform and deep learning approach
title_sort estimation of lower limb torque a novel hybrid method based on continuous wavelet transform and deep learning approach
topic Inertial measurement units (IMU)
Lower limb joint torque
Deep learning approach
Bi-LSTM
Continuous wavelet transform (CWT)
url https://peerj.com/articles/cs-2888.pdf
work_keys_str_mv AT shuxu estimationoflowerlimbtorqueanovelhybridmethodbasedoncontinuouswavelettransformanddeeplearningapproach
AT taowang estimationoflowerlimbtorqueanovelhybridmethodbasedoncontinuouswavelettransformanddeeplearningapproach
AT zenghuiding estimationoflowerlimbtorqueanovelhybridmethodbasedoncontinuouswavelettransformanddeeplearningapproach
AT yuwang estimationoflowerlimbtorqueanovelhybridmethodbasedoncontinuouswavelettransformanddeeplearningapproach
AT tongshengwan estimationoflowerlimbtorqueanovelhybridmethodbasedoncontinuouswavelettransformanddeeplearningapproach
AT dezhangxu estimationoflowerlimbtorqueanovelhybridmethodbasedoncontinuouswavelettransformanddeeplearningapproach
AT xianjunyang estimationoflowerlimbtorqueanovelhybridmethodbasedoncontinuouswavelettransformanddeeplearningapproach
AT tingsun estimationoflowerlimbtorqueanovelhybridmethodbasedoncontinuouswavelettransformanddeeplearningapproach
AT mengli estimationoflowerlimbtorqueanovelhybridmethodbasedoncontinuouswavelettransformanddeeplearningapproach