Search

Introduction: Children’s walking patterns evolve with age, exhibiting less repetitiveness at a young age and more variability than adults. Three-dimensional gait analysis (3DGA) is crucial for understanding and treating lower limb movement disorders in children, traditionally performed using Optical Motion Capture (OMC). Inertial Measurement Units (IMUs) offer a cost-effective alternative to OMC, although challenges like drift errors persist. Machine learning (ML) models can mitigate these issues in adults, prompting an investigation into their applicability to a heterogeneous pediatric population. This study aimed at 1) quantifying personalized and generalized ML models’ performance for predicting gait time series in typically developed (TD) children using IMUs data, 2) Comparing random forest (RF) and convolutional neural networks (CNN) models’ performance, 3) Finding the optimal number of IMUs required for accurate predictions.Methodology: Seventeen TD children, aged 6 to 15, participated in data collection involving OMC, force plates, and IMU sensors. Joint kinematics and kinetics (targets) were computed from OMC and force plates’ data using OpenSim. Tsfresh, a Python package, extracted features from raw IMU data. Each target’s ten most important features were input in the development of personalized and generalized RF and CNN models. This procedure was initially conducted with 7 IMUs placed on all lower limb segments and then performed using only two IMUs on the feet.Results: Findings suggested that the RF and CNN models demonstrated comparable performance. RF predicted joint kinematics with a 9.5% and 19.9% NRMSE for personalized and generalized models, respectively, and joint kinetics with an NRMSE of 10.7% for personalized and 15.2% for generalized models in TD children. Personalized models provided accurate estimations from IMU data in children, while generalized models lacked accuracy due to the limited dataset. Furthermore, reducing the number of IMUs from 7 to 2 did not affect the results, and the performance remained consistent.Discussion: This study proposed a promising personalized approach for gait time series prediction in children, involving an RF model and two IMUs on the feet.

The development of fatigue elicits multiple adaptations from the neuromuscular system. Muscle synergies are common patterns of neuromuscular activation that have been proposed as the building blocks of human movement. We wanted to identify possible adaptations of muscle synergies to the development of fatigue in the upper limb. Recent studies have reported that synergy structure remains invariant during the development of fatigue, but these studies did not examine isolated synergies. We propose a novel approach to characterise synergy adaptations to fatigue by taking advantage of the spatial tuning of synergies. This approach allows improved identification of changes to individual synergies that might otherwise be confounded by changing contributions of overlapping synergies. To analyse upper limb synergies, we applied non-negative matrix factorization to 14 EMG signals from muscles of 11 participants performing isometric contractions. A preliminary multidirectional task was used to identify synergy directional tuning. A subsequent fatiguing task was designed to fatigue the participants in their synergies’ preferred directions. Both tasks provided virtual reality feedback of the applied force direction and magnitude, and were performed at 40% of each participant’s maximal voluntary force. Five epochs were analysed throughout the fatiguing task to identify progressive changes of EMG amplitude, median frequency, synergy structure, and activation coefficients. Three to four synergies were sufficient to account for the variability contained in the original data. Synergy structure was conserved with fatigue, but interestingly synergy activation coefficients decreased on average by 24.5% with fatigue development. EMG amplitude did not change systematically with fatigue, whereas EMG median frequency consistently decreased across all muscles. These results support the notion of a neuromuscular modular organisation as the building blocks of human movement, with adaptations to synergy recruitment occurring with fatigue. When synergy tuning properties are considered, the reduction of activation of muscle synergies may be a reliable marker to identify fatigue.

Introduction: Children's walking patterns evolve with age, exhibiting less repetitiveness at a young age and more variability than adults. Three-dimensional gait analysis (3DGA) is crucial for understanding and treating lower limb movement disorders in children, traditionally performed using Optical Motion Capture (OMC). Inertial Measurement Units (IMUs) offer a cost-effective alternative to OMC, although challenges like drift errors persist. Machine learning (ML) models can mitigate these issues in adults, prompting an investigation into their applicability to a heterogeneous pediatric population. This study aimed at 1) quantifying personalized and generalized ML models' performance for predicting gait time series in typically developed (TD) children using IMUs data, 2) Comparing random forest (RF) and convolutional neural networks (CNN) models' performance, 3) Finding the optimal number of IMUs required for accurate predictions. Methodology: Seventeen TD children, aged 6 to 15, participated in data collection involving OMC, force plates, and IMU sensors. Joint kinematics and kinetics (targets) were computed from OMC and force plates' data using OpenSim. Tsfresh, a Python package, extracted features from raw IMU data. Each target's ten most important features were input in the development of personalized and generalized RF and CNN models. This procedure was initially conducted with 7 IMUs placed on all lower limb segments and then performed using only two IMUs on the feet. Results: Findings suggested that the RF and CNN models demonstrated comparable performance. RF predicted joint kinematics with a 9.5% and 19.9% NRMSE for personalized and generalized models, respectively, and joint kinetics with an NRMSE of 10.7% for personalized and 15.2% for generalized models in TD children. Personalized models provided accurate estimations from IMU data in children, while generalized models lacked accuracy due to the limited dataset. Furthermore, reducing the number of IMUs from 7 to 2 did not affect the results, and the performance remained consistent. Discussion: This study proposed a promising personalized approach for gait time series prediction in children, involving an RF model and two IMUs on the feet., Competing Interests: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest., (Copyright © 2024 Mohammadi Moghadam, Ortega Auriol, Yeung and Choisne.)

When reconstructing bone geometry to calculate joint kinematics, shape-model scaling can be more accurate and repeatable than linear scaling given the same anatomical landmarks. This study perturbed anatomical landmarks from optical motion capture and determined the robustness of shape-model scaling to misplaced markers compared to a traditional approach of linear scaling. We hypothesised that shape-model scaling would be less susceptible to variance in marker positions compared to linear scaling. The positions of hip joint centres and femoral/tibial segment lengths across perturbations were compared to determine each scaling method's range of geometric variation. The standard deviation (SD) of the hip joint centre location from the shape model had a maximum of 1.4 mm, compared to 4.2 mm for linear scaling. Femoral and tibial segments displayed SD's of 5.4 mm and 5.2 mm when shape-model scaled, compared to 9.2 mm and 9.5 mm with linear scaling, respectively, thus supporting our hypothesis. Geometric constraints within a shape model provide robustness to marker misplacement providing potential improvements in repeatability and data exchange., Competing Interests: Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2023. Published by Elsevier Ltd.)

Context: Pathologic plantar flexion frequently occurs after operative repair of the Achilles tendon (AT) because of immobilization and non-weight bearing in the first weeks of traditional rehabilitation. Novel rehabilitation strategies that apply mobilization and weight bearing have been proposed, but their effects on medial gastrocnemius myotendinous junction displacement (MJD) and isometric plantar-flexion strength (PFS) are unknown., Objective: To compare the effects of 12 weeks of immediate versus traditional rehabilitation on MJD and PFS in patients with percutaneous AT repair and to compare AT rupture scores (ATRSs) during follow-up., Design: Controlled laboratory study., Setting: Human performance laboratory., Patients or Other Participants: A total of 26 amateur soccer players (age = 42.3 ± 9.7 years, body mass index = 29.5 ± 3.9 kg/m 2 ) with percutaneous AT repair., Intervention(s): Athletes were randomly divided into 2 groups: an immediate group, given physical therapy from day 1 to day 84, and a traditional group, given physical therapy from day 29 to day 84. We used repeated-measures analysis of variance to compare the data., Main Outcome Measure(s): We measured MJD and PFS at days 28 (fourth week), 56 (eighth week), and 84 (12th week) after AT repair., Results: After 12 weeks of rehabilitation, we observed a large clinically meaningful effect and statistical difference between groups. At day 28, the immediate group showed higher values for PFS (P = .002), MJD (P = .02), and ATRS (P = .002) than the traditional group. At day 56, the immediate group presented higher values for MJD (P = .02) and ATRS (P = .009). At day 84, the immediate group registered more MJD (P = .001)., Conclusions: Compared with traditional rehabilitation, 12 weeks of immediate rehabilitation after percutaneous AT repair resulted in better MJD, PFS, and ATRS after 4 weeks; better MJD and ATRS after 8 weeks; and better MJD after 12 weeks.