Jongsang Son

Falls are a major cause of injury in older adults. Although bending the knees during a fall has been shown to reduce stress on the hip, knee motion during falls is not well understood because laboratory fall studies are limited by safety concerns and marker occlusion in motion capture systems. AI-based pose estimation may help overcome these challenges, but its accuracy in measuring joint angles during falls has not yet been validated. We evaluated three pose estimation models (OpenPose, VideoPose3D, WHAM) for analyzing knee kinematics in video-captured falls. A total of 121 videos of 13 older adults (64.0 ± 5.9 years) falling sideways, utilizing diverse fall strategies (knee block, stick-like, tuck-and-roll), in a lab setting were analyzed. Each model generated time series of knee angles from the videos, from which knee flexion angles at ground impact were calculated and compared to ground truth data from a motion capture system. Agreement with the ground truth was assessed using mean absolute error (MAE), mean absolute percentage error (MAPE), and bias, analyzed across viewing planes (sagittal vs. frontal) and leg sides (impact vs. opposite). WHAM demonstrated the highest accuracy (MAPE:13.61 ± 10.55%) with minimal bias (<10%), consistently performing well across all views and leg sides. OpenPose performed similar to WHAM in the sagittal view (MAPE:14.38 ± 9.63%) but poorly in the frontal view (MAPE:71.33 ± 17.24%) due to substantial underestimation (bias:-71.33 ± 17.24%). VideoPose3D showed poor accuracy across all conditions (MAPE:39.09 ± 20.54%). WHAM also characterized differences in knee flexion kinematics between fall strategies (e.g., least vs. most knee flexion) but did not fully reproduce side-specific kinematic differences between legs, particularly for tuck-and-roll falls. This is the first study to validate pose estimation algorithms for estimating knee impact angles from video-captured falls in older adults. Future work should fine-tune WHAM using fall-specific data to further improve its performance in tracking body movements during falls.

The purpose of this study is to develop a deep neural network model for predicting soleus muscle activation patterns from time-series lower-limb joint angles collected during level ground walking at different speeds and to evaluate the prediction performance. An open dataset was used to obtain full lower-limb kinematics, including pelvis, hip, knee, and ankle joints, and soleus muscle activation patterns from 20 control adults. Long short-term memory (LSTM)-based deep learning models were developed and then evaluated for prediction performance by conducting both the random split cross-validation (CVM1) and the leave-one-subject-out cross-validation (CVM2). For both cross-validation methods, the developed models yielded promising error and regression metrics such as the root mean square error and coefficient of determination. However, the CVM2 demonstrated that the prediction performance can be sensitive to individual datasets. The subject factors, such as age, sex, and walking speed, appear to have a negligible effect on the prediction performance for the CVM1. This study demonstrated the feasibility of the developed models to be a template for a potential tool that quantifies muscle activation patterns from joint angles during level ground walking at different speeds.

Myographic signals can effectively detect and assess subtle changes in muscle function; however, their measurement and analysis are often limited in clinical settings compared to inertial measurement units. Recently, the advent of artificial intelligence (AI) has made the analysis of complex myographic signals more feasible. This scoping review aims to examine the use of myographic signals in conjunction with AI for assessing motor impairments and highlight potential limitations and future directions. We conducted a systematic search using specific keywords in the Scopus and PubMed databases. After a thorough screening process, 111 relevant studies were selected for review. These studies were organized based on target applications (measurement modality, measurement location, and AI application task), sample demographics (age, sex, ethnicity, and pathology), and AI models (general approach and algorithm type). Among various myographic measurement modalities, surface electromyography was the most commonly used. In terms of AI approaches, machine learning with feature engineering was the predominant method, with classification tasks being the most common application of AI. Our review also noted a significant bias in participant demographics, with a greater representation of males compared to females and healthy individuals compared to clinical populations. Overall, our findings suggest that integrating myographic signals with AI has the potential to provide more objective and clinically relevant assessments of motor impairments.

This study explored the impact of a four-week Neuromuscular Electrical Stimulation (NMES)-induced eccentric contraction training on single-leg standing balance and muscle strength in 17 healthy adults. The unique training approach involved active antagonist muscle contraction during NMES. Post-training results revealed significant improvements in balance, with notable reductions in Center of Pressure (CoP) trajectory velocity (mean reduction: 0.07 ± 0.01 cm/s, p < 0.05) and range (mean reduction: 2.98 ± 0.53 cm, p < 0.05) on a firm surface. While increases in dorsiflexion force (mean increase: 21.43 ± 0.79 N, p < 0.05) and muscle activation were observed, these were not statistically significant. Changes in muscle pennation angles were also not significant (mean change: 0.43 ± 0.06 degrees, p > 0.05), underscoring the complexity of muscle adaptation processes. This study highlights NMES’s potential in enhancing balance and proprioceptive sensing, suggesting its promising applications in neuromuscular rehabilitation. However, further research is needed to fully understand its impact.

Myographic signals can effectively detect and assess subtle changes in muscle function; however, their measurement and analysis are often limited in clinical settings compared to inertial measurement units (IMUs). Recently, the advent of artificial intelligence (AI) has made the analysis of complex myographic signals more feasible. This scoping review aims to examine the use of myographic signals in conjunction with AI for assessing motor impairments, while also highlighting potential limitations and future directions. We conducted a systematic search using specific keywords in the Scopus and PubMed databases. After a thorough screening process, 111 relevant studies were selected for review. These studies were organized based on target applications (measurement modality, modality location, and AI model purpose/application), sample demographics (age, sex, ethnicity, and pathology), and AI models (general approach and algorithm type). Among the various myographic measurement modalities, surface electromyography (sEMG) was the most commonly used. In terms of AI approaches, machine learning with feature engineering was the predominant method, with classification tasks being the most common application of AI. Our review also noted a significant bias in participant demographics, with a greater representation of males compared to females and healthy individuals compared to patient groups. Overall, our findings suggest that integrating myographic signals with AI has the potential to provide more objective and clinically relevant assessments of motor impairments.

During voluntary isometric contractions of upper extremity muscles in individuals with chronic stroke, the magnitude of force variability appears to increase consistently as force increases. However, research on how such force variability changes with increasing motor impairment remains limited. This study aims to determine whether force variability is increased on the paretic side during either index finger abduction or elbow flexion in the same group of stroke survivors, and whether these changes are consistent across different muscles. Force variability was assessed using the standard deviation of force during sustained isometric contractions. Linear mixed-effects models were implemented to test whether force variability is changed on the paretic side post stroke, and whether such alterations show dependence on force level and on the degree of impairment. The results demonstrated a significant increase in force variability on the paretic side across force levels during finger abduction, while force variability for elbow flexion was increased only at high force levels. In addition, the force variability appears to increase as isometric elbow flexion force increases, whereas no clear trend was found during index finger abduction. The increase in force variability demonstrated moderate-strong dependence on the reduction in maximum muscle strength on the paretic side during elbow flexion, suggesting that monitoring force variability could potentially serve as a quantitative diagnostic tool for assessing severity of impairment in motor control and for raising potential mechanisms at the motor unit level.

Quantifying muscle strength is an important measure in clinical settings; however, there is a lack of practical tools that can be deployed for routine assessment. The purpose of this study is to propose a deep learning model for ankle plantar flexion torque prediction from time-series mechanomyogram (MMG) signals recorded during isometric contractions (i.e., a similar form to manual muscle testing procedure in clinical practice) and to evaluate its performance. Four different deep learning models in terms of model architecture (based on a stacked bidirectional long short-term memory and dense layers) were designed with different combinations of the number of units (from 32 to 512) and dropout ratio (from 0.0 to 0.8), and then evaluated for prediction performance by conducting the leave-one-subject-out cross-validation method from the 10-subject dataset. As a result, the models explained more variance in the untrained test dataset as the error metrics (e.g., root-mean-square error) decreased and as the slope of the relationship between the measured and predicted joint torques became closer to 1.0. Although the slope estimates appear to be sensitive to an individual dataset, >70% of the variance in nine out of 10 datasets was explained by the optimal model. These results demonstrated the feasibility of the proposed model as a potential tool to quantify average joint torque during a sustained isometric contraction.

A skeletal muscle's peak force production and excursion are based on its architectural properties that are, in turn, determined by its mass, muscle fiber length and physiological cross-sectional area (PCSA). In the classic interspecific study of mammalian muscle scaling, it was demonstrated that muscle mass scales positively allometrically with body mass whereas fiber length scales isometrically with body mass, indicating that larger mammals have stronger leg muscles than they would if they were geometrically similar to smaller ones. Although this relationship is highly significant across species, there has never been a detailed intraspecific architectural scaling study. We have thus created a large dataset of 896 muscles across 34 human lower extremities (18 females and 16 males) with a size range including approximately 90% and 70% of the United States population height and mass, respectively, across the range 36–103 years. Our purpose was to quantify the scaling relationships between human muscle architectural properties and body size. We found that human muscles depart greatly from isometric scaling because muscle mass scales with body mass1.3 (larger exponent than isometric scaling of 1.0) and muscle fiber length scales with negative allometry with body mass0.1 (smaller exponent than isometric scaling of 0.33). Based on the known relationship between architecture and function, these results suggest that human muscles place a premium on muscle force production (mass and PCSA) at the expense of muscle excursion (fiber length) with increasing body size, which has implications for understanding human muscle design as well as biomechanical modeling.

Editorial: Neuromuscular adaptations to sensorimotor stimulation protocols: potential rehabilitative interventions for individuals with central or peripheral neuromuscular injuries

The aim of this review was to provide an overview of the recent findings on the acute and chronic effects of static stretching on joint behaviors and neuromuscular responses and to discuss the overall effects of acute and chronic static stretching on selected outcomes via meta-analyses, using a total of 50 recent studies. The results of our meta-analyses demonstrated that acute static stretching results in increased range of motion (ROM), decreased passive resistive torque (PRT), increased maximum tolerable PRT (PRT_max), decreased maximum voluntary isometric torque, decreased muscle–tendon unit stiffness, decreased muscle stiffness, decreased tendon stiffness, and decreased shear elastic modulus. Moreover, the chronic effects of static stretching included increased ROM, increased PRT_max, decreased muscle stiffness, and decreased shear elastic modulus (or shear wave speed). These results suggest that static stretching interventions have the potential to increase ROM and reduce the mechanical properties of muscle–tendon tissue, but they may not change corticospinal excitability and spinal reflex excitability or muscle architecture parameters.