Human kinematic, kinetic and EMG data during different walking and stair ascending and descending tasks

Marker Data Enhancement For Markerless Motion Capture

Objective

Human pose estimation models can measure movement from videos at a large scale and low cost; however, open-source pose estimation models typically detect only sparse keypoints, which leads to inaccurate joint kinematics. OpenCap, a freely available service for researchers to measure movement from videos, addresses this issue using a deep learning model- the marker enhancer-that transforms sparse keypoints into dense anatomical markers. However, OpenCap performs poorly on movements not included in the training data. Here, we create a much larger and more diverse training dataset and develop a more accurate and generalizable marker enhancer.

Methods

We compiled marker-based motion capture data from 1176 subjects and synthesized 1433 hours of keypoints and anatomical markers to train the marker enhancer. We evaluated its accuracy in computing kinematics using both benchmark movement videos and synthetic data representing unseen, diverse movements.

Results

The marker enhancer improved kinematic accuracy on benchmark movements (mean error: 4.1°, max: 8.7°) compared to using video keypoints (mean: 9.6°, max: 43.1°) and OpenCap's original enhancer (mean: 5.3°, max: 11.5°). It also better generalized to unseen, diverse movements (mean: 4.1°, max: 6.7°) than OpenCap's original enhancer (mean: 40.4°, max: 252.0°).

Conclusion

Our marker enhancer demonstrates both accuracy and generalizability across diverse movements.

Significance

We integrated the marker enhancer into OpenCap, thereby offering its thousands of users more accurate measurements across a broader range of movements.

Knee Angle Estimation from Surface EMG during Walking Using Attention-Based Deep Recurrent Neural Networks: Feasibility and Initial Demonstration in Cerebral Palsy

Accurately estimating knee joint angle during walking from surface electromyography (sEMG) signals can enable more natural control of wearable robotics like exoskeletons. However, challenges exist due to variability across individuals and sessions. This study evaluates an attention-based deep recurrent neural network combining gated recurrent units (GRUs) and an attention mechanism (AM) for knee angle estimation. Three experiments were conducted. First, the GRU-AM model was tested on four healthy adolescents, demonstrating improved estimation compared to GRU alone. A sensitivity analysis revealed that the key contributing muscles were the knee flexor and extensors, highlighting the ability of the AM to focus on the most salient inputs. Second, transfer learning was shown by pretraining the model on an open source dataset before additional training and testing on the four adolescents. Third, the model was progressively adapted over three sessions for one child with cerebral palsy (CP). The GRU-AM model demonstrated robust knee angle estimation across participants with healthy participants (mean RMSE 7 degrees) and participants with CP (RMSE 37 degrees). Further, estimation accuracy improved by 14 degrees on average across successive sessions of walking in the child with CP. These results demonstrate the feasibility of using attention-based deep networks for joint angle estimation in adolescents and clinical populations and support their further development for deployment in wearable robotics.

A biomechanics and energetics dataset of neurotypical adults walking with and without kinematic constraints

Numerous studies have explored the biomechanics and energetics of human walking, offering valuable insights into how we walk. However, prior studies focused on changing external factors (e.g., walking speed) and examined group averages and trends rather than individual adaptations in the presence of internal constraints (e.g., injury-related muscle weakness). To address this gap, this paper presents an open dataset of human walking biomechanics and energetics collected from 21 neurotypical young adults. To investigate the effects of internal constraints (reduced joint range of motion), the participants are both the control group (free walking) and the intervention group (constrained walking - left knee fully extended using a passive orthosis). Each subject walked on a dual-belt treadmill at three speeds (0.4, 0.8, and 1.1 m/s) and five step frequencies ( - 10% to 20% of their preferred frequency) for a total of 30 test conditions. The dataset includes raw and segmented data featuring ground reaction forces, joint motion, muscle activity, and metabolic data. Additionally, a sample code is provided for basic data manipulation and visualisation.

3D motion analysis dataset of healthy young adult volunteers walking and running on overground and treadmill

Used on clinical and sportive context, three-dimensional motion analysis is considered as the gold standard in the biomechanics field. The proposed dataset has been established on 30 asymptomatic young participants. Volunteers were asked to walk at slow, comfortable and fast speeds, and to run at comfortable and fast speeds on overground and treadmill using shoes. Three dimensional trajectories of 63 reflective markers, 3D ground reaction forces and moments were simultaneously recorded. A total of 4840 and 18159 gait cycles were measured for overground and treadmill walking, respectively. Additionally, 2931 and 18945 cycles were measured for overground and treadmill running, respectively. The dataset is presented in C3D and CSV files either in raw or pre-processed format. The aim of this dataset is to provide a complete set of data that will help for the gait characterization during clinical gait analysis and in a sportive context. This data could be used for the creation of a baseline database for clinical purposes to research activities exploring the gait and the run.

Continuous Locomotion Mode and Task Identification for an Assistive Exoskeleton Based on Neuromuscular-Mechanical Fusion

Human walking parameters exhibit significant variability depending on the terrain, speed, and load. Assistive exoskeletons currently focus on the recognition of locomotion terrain, ignoring the identification of locomotion tasks, which are also essential for control strategies. The aim of this study was to develop an interface for locomotion mode and task identification based on a neuromuscular-mechanical fusion algorithm. The modes of level and incline and tasks of speed and load were explored, and seven able-bodied participants were recruited. A continuous stream of assistive decisions supporting timely exoskeleton control was achieved according to the classification of locomotion. We investigated the optimal algorithm, feature set, window increment, window length, and robustness for precise identification and synchronization between exoskeleton assistive force and human limb movements (human-machine collaboration). The best recognition results were obtained when using a support vector machine, a root mean square/waveform length/acceleration feature set, a window length of 170, and a window increment of 20. The average identification accuracy reached 98.7% ± 1.3%. These results suggest that the surface electromyography-acceleration can be effectively used for locomotion mode and task identification. This study contributes to the development of locomotion mode and task recognition as well as exoskeleton control for seamless transitions.

On the relation between gait speed and gait cycle duration for walking on even ground

Gait models and reference motions are essential for the objective assessment of walking patterns and therapy progress, as well as research in the field of wearable robotics and rehabilitation devices in general. A human can achieve a desired gait speed by adjusting stride length and/or stride frequency. It is hypothesized that sex, age, and physique of a person have a significant influence on the combination of these parameters. A mathematical description of the relation between gait speed and its determinants is presented in the form of a parameterized analytic function. Based on the statistical significance of the parameters, three models are derived. The first two models are valid for slow to fast walking, which is defined as the interval of approximately 0.6-2.0ms-1, assuming a linear relation of gait speed and stride length, and a non-linear relation of gait speed and stride duration, respectively. The third model is valid for a defined range of walking speed centered at a certain (preferred or spontaneous) gait speed. The latter assumes a constant walk ratio, i.e. the ratio between step or stride length and step or stride frequency, and is recommended for walking at a speed of 1.0-1.6ms-1. On the basis of a large pool of gait datasets, regression coefficients with significance for age and/or body mass index are identified. The presented models allow to estimate the gait cycle duration based on gait speed, sex, age and body mass index of healthy persons walking on even ground.

A human lower-limb biomechanics and wearable sensors dataset during cyclic and non-cyclic activities

Tasks of daily living are often sporadic, highly variable, and asymmetric. Analyzing these real-world non-cyclic activities is integral for expanding the applicability of exoskeletons, protheses, wearable sensing, and activity classification to real life, and could provide new insights into human biomechanics. Yet, currently available biomechanics datasets focus on either highly consistent, continuous, and symmetric activities, such as walking and running, or only a single specific non-cyclic task. To capture a more holistic picture of lower limb movements in everyday life, we collected data from 12 participants performing 20 non-cyclic activities (e.g. sit-to-stand, jumping, squatting, lunging, cutting) as well as 11 cyclic activities (e.g. walking, running) while kinematics (motion capture and IMUs), kinetics (force plates), and electromyography (EMG) were collected. This dataset provides normative biomechanics for a highly diverse range of activities and common tasks from a consistent set of participants and sensors.

A full-body motion capture gait dataset of 138 able-bodied adults across the life span and 50 stroke survivors

This reference dataset contains biomechanical data of 138 able-bodied adults (21-86 years) and 50 stroke survivors walking bare-footed at their preferred speed. It is unique due to its size, and population, including adults across the life-span and over 70 years, as well as stroke survivors. Full-body kinematics (PiG-model), kinetics and muscle activity of 14 back and lower limbs muscles was collected with a Vicon motion capture system, ground-embedded force plates, and a synchronized surface EMG system. The data is reliable to compare within and between groups as the same methodology and infrastructure were used to gather all data. Both source files (C3D) and post-processed ready-to-use stride-normalized kinematics, kinetics and EMG data (MAT-file, Excel file) are available, allowing high flexibility and accessibility of analysis for both researchers and clinicians. These records are valuable to examine ageing, typical and hemiplegic gait, while also offering a wide range of reference data which can be utilized for age-matched controls during normal walking.

High-density EMG, IMU, kinetic, and kinematic open-source data for comprehensive locomotion activities

Novel sensor technology enables new insights in the neuromechanics of human locomotion that were previously not possible. Here, we provide a dataset of high-density surface electromyography (HDsEMG) and high-resolution inertial measurement unit (IMU) signals, along with motion capture and force data for the lower limb of 10 healthy adults during multiple locomotion modes. The participants performed level-ground and slope walking, as well as stairs ascent/descent, side stepping gait, and stand-to-walk and sit-to-stand-to-walk, at multiple walking speeds. These data can be used for the development and validation of locomotion mode recognition and control algorithms for prosthetics, exoskeletons, and bipedal robots, and for motor control investigations.

Inertial measurement data from loose clothing worn on the lower body during everyday activities

Embedding sensors into clothing is promising as a way for people to wear multiple sensors easily, for applications such as long-term activity monitoring. To our knowledge, this is the first published dataset collected from sensors in loose clothing. 6 Inertial Measurement Units (IMUs) were configured as a 'sensor string' and attached to casual trousers such that there were three sensors on each leg near the waist, thigh, and ankle/lower-shank. Participants also wore an Actigraph accelerometer on their dominant wrist. The dataset consists of 15 participant-days worth of data collected from 5 healthy adults (age range: 28-48 years, 3 males and 2 females). Each participant wore the clothes with sensors for between 1 and 4 days for 5-8 hours per day. Each day, data were collected while participants completed a fixed circuit of activities (with a video ground truth) as well as during free day-to-day activities (with a diary). This dataset can be used to analyse human movements, transitional movements, and postural changes based on a range of features.

Characterization of the ankle dynamic joint stiffness as a function of gait speed for overground and treadmill walking

The ankle dynamic joint stiffness (DJS), defined as the slope of the joint angle-moment plot, measures the resistance of the ankle joint to movement when the foot is in contact with the ground. DJS helps to stabilize the ankle joint, and its characterization helps to identify gait pathology and assist foot prosthesis design. This study analyzes the available gait dynamics data to obtain ankle DJS parameters for population groups according to age, gender, and gait speed for overground and treadmill walking. This study classified the groups into five walking speeds normalized using the Froude number. Herein, 12 ankle DJS parameters were determined. These include four linear segments: controlled plantar flexion (CP), early response phase (ERP), large response phase (LRP), and descending phase (DP), their corresponding turning points, the net mechanical work, the absorbed work, and the loop direction. Ankle dynamics data for 92 individuals was collected from two gait data repositories. The analysis reveals a notable disparity in stiffness values between overground and treadmill gait. Specifically, the CP stiffness is significantly higher for overground gait. In contrast, the DP stiffness displays an opposing pattern, with higher values observed during treadmill walking. A negative stiffness for LRP was found at fast speeds for all groups. The sorted data, analysis tools, and findings of this study are meant to help practitioners design prosthetic and rehabilitation devices based on age, gender, and walking environment at different gait speeds.

Estimating vertical ground reaction forces during gait from lower limb kinematics and vertical acceleration using wearable inertial sensors

One of the most important forces generated during gait is the vertical ground reaction force (vGRF). This force can be measured using force plates, but these can limit the scope of gait analysis. This paper presents a method to estimate the vGRF using inertial measurement units (IMU) and machine learning techniques. Four wearable IMUs were used to obtain flexion/extension angles of the hip, knee, and ankle joints, and an IMU placed over the C7 vertebra to measure vertical acceleration. We trained and compared the performance of two machine learning algorithms: feedforward neural networks (FNN) and random forest (RF). We investigated the importance of the inputs introduced into the models and analyzed in detail the contribution of lower limb kinematics and vertical acceleration to model performance. The results suggest that the inclusion of vertical acceleration increases the root mean square error in the FNN, while the RF appears to decrease it. We also analyzed the ability of the models to construct the force signal, with particular emphasis on the magnitude and timing of the vGRF peaks. Using the proposed method, we concluded that FNN and RF models can estimate the vGRF with high accuracy.

Attention-Based Deep Recurrent Neural Network to Estimate Knee Angle During Walking from Lower-Limb EMG

Accurate prediction of joint angle during walking from surface electromyography (sEMG) offers the potential to infer movement intention and therefore represents a potentially useful approach for adaptive control of wearable robotics. Here, we present the use of a recurrent neural network (RNN) with gated recurrent units (GRUs) and an attention mechanism to estimate knee angle during overground walking from sEMG and its initial offline validation in healthy adolescents. Our results show that the attention mechanism improved estimation accuracy by focusing on the most relevant parts of the input dataset within each time window, particularly muscles active during knee excursion. Sensitivity analysis revealed knee extensor and flexor muscles to be most salient in accurately estimating joint angle. Additionally, we demonstrate the ability of the GRU-RNN approach to accurately estimate knee angle during overground walking in a child with cerebral palsy (CP) in the presence of exoskeleton knee extension assistance. Collectively, our findings establish the initial feasibility of using this approach to estimate user movement from sEMG, which is particularly important for developing robotic exoskeletons for children with neuromuscular disorders such as CP.

Clinical Gait Analysis: Characterizing Normal Gait and Pathological Deviations Due to Neurological Diseases

This study addresses the characterization of normal gait and pathological deviations induced by neurological diseases, considering knee angular kinematics in the sagittal plane. We propose an unsupervised approach based on Dynamic Time Warping (DTW) to identify different normal gait profiles (NGPs) corresponding to real cycles representing the overall behavior of healthy subjects, instead of considering an average reference, as done in the literature. The obtained NGPs are then used to measure the deviations of pathological gait cycles from normal gait with DTW. Hierarchical Clustering is applied to stratify deviations into clusters. Results show that three NGPs are necessary to finely characterize the heterogeneity of normal gait and accurately quantify pathological deviations. In particular, we automatically identify which lower limb is affected for Hemiplegic patients and characterize the severity of motor impairment for Paraplegic patients. Concerning Tetraplegic patients, different profiles appear in terms of impairment severity. These promising results are obtained by considering the raw description of gait signals. Indeed, we have shown that normalizing signals removes the temporal properties of signals, inducing a loss of dynamic information that is crucial for accurately measuring pathological deviations. Our methodology could be exploited to quantify the impact of therapies on gait rehabilitation.

Database of lower limb kinematics and electromyography during gait-related activities in able-bodied subjects

This data descriptor describes the Roessingh Research & Development-MyLeg database for activity prediction (MyPredict), containing three data sets. These data sets contain data from 55 able-bodied subjects, mean age 24 ± 2 years, measured in 85 measurement sessions. Measurement sessions consisted of trials containing sitting, standing, overground walking, stair ascent, stair descent, ramp ascent, ramp descent, walking on uneven terrain and walking in simulated confined spaces. Subjects were measured using eight inertial measurement units in combination with different types of sEMG. Recorded kinematics consisted of joint angles, sensor accelerations, angular velocity, orientation and virtual marker positions. sEMG was recorded using bipolar sEMG, multi-array sEMG or a combination of both. All data showed excellent correlation with other online available data sets. The data reported in this descriptor forms a solid basis for research into myoelectric pattern recognition, myoelectric control development and electromyography to be used in data-driven applications.

The use of nonnormalized surface EMG and feature inputs for LSTM-based powered ankle prosthesis control algorithm development

Advancements in instrumentation support improved powered ankle prostheses hardware development. However, control algorithms have limitations regarding number and type of sensors utilized and achieving autonomous adaptation, which is key to a natural ambulation. Surface electromyogram (sEMG) sensors are promising. With a minimized number of sEMG inputs an economic control algorithm can be developed, whereas limiting the use of lower leg muscles will provide a practical algorithm for both ankle disarticulation and transtibial amputation. To determine appropriate sensor combinations, a systematic assessment of the predictive success of variations of multiple sEMG inputs in estimating ankle position and moment has to conducted. More importantly, tackling the use of nonnormalized sEMG data in such algorithm development to overcome processing complexities in real-time is essential, but lacking. We used healthy population level walking data to (1) develop sagittal ankle position and moment predicting algorithms using nonnormalized sEMG, and (2) rank all muscle combinations based on success to determine economic and practical algorithms. Eight lower extremity muscles were studied as sEMG inputs to a long-short-term memory (LSTM) neural network architecture: tibialis anterior (TA), soleus (SO), medial gastrocnemius (MG), peroneus longus (PL), rectus femoris (RF), vastus medialis (VM), biceps femoris (BF) and gluteus maximus (GMax). Five features extracted from nonnormalized sEMG amplitudes were used: integrated EMG (IEMG), mean absolute value (MAV), Willison amplitude (WAMP), root mean square (RMS) and waveform length (WL). Muscle and feature combination variations were ranked using Pearson's correlation coefficient (r > 0.90 indicates successful correlations), the root-mean-square error and one-dimensional statistical parametric mapping between the original data and LSTM response. The results showed that IEMG+WL yields the best feature combination performance. The best performing variation was MG + RF + VM (r_position = 0.9099 and r_moment = 0.9707) whereas, PL (r_position = 0.9001, r_moment = 0.9703) and GMax+VM (r_position = 0.9010, r_moment = 0.9718) were distinguished as the economic and practical variations, respectively. The study established for the first time the use of nonnormalized sEMG in control algorithm development for level walking.

Surface electromyogram, kinematic, and kinetic dataset of lower limb walking for movement intent recognition

Surface electromyogram (sEMG) offers a rich set of motor information for decoding limb motion intention that serves as a control input to Intelligent human-machine synergy systems (IHMSS). Despite growing interest in IHMSS, the current publicly available datasets are limited and can hardly meet the growing demands of researchers. This study presents a novel lower limb motion dataset (designated as SIAT-LLMD), comprising sEMG, kinematic, and kinetic data with corresponding labels acquired from 40 healthy humans during 16 movements. The kinematic and kinetic data were collected using a motion capture system and six-dimensional force platforms and processed using OpenSim software. The sEMG data were recorded using nine wireless sensors placed on the subjects' thigh and calf muscles on the left limb. Besides, SIAT-LLMD provides labels to classify the different movements and different gait phases. Analysis of the dataset verified the synchronization and reproducibility, and codes for effective data processing are provided. The proposed dataset can serve as a new resource for exploring novel algorithms and models for characterizing lower limb movements.

An Integrated Analog Front-End System on Flexible Substrate for the Acquisition of Bio-Potential Signals

The application of a versatile, low-temperature thin-film transistor (TFT) technology is presently described as the implementation on a flexible substrate of an analog front-end (AFE) system for the acquisition of bio-potential signals. The technology is based on semiconducting amorphous indium-gallium-zinc oxide (IGZO). The AFE system consists of three monolithically integrated constituent components: a bias-filter circuit with a bio-compatible low cut-off frequency of ≈1 Hz, a 4-stage differential amplifier offering a large gain-bandwidth product of ≈955 kHz, and an additional notch filter exhibiting over 30 dB suppression of the power-line noise. Respectively built using conductive IGZO electrodes with thermally induced donor agents and enhancement-mode fluorinated IGZO TFTs with exceptionally low leakage current, both capacitors and resistors with significantly reduced footprints are realized. Defined as the ratio of the gain-bandwidth product of an AFE system to its area, a record-setting figure-of-merit of ≈86 kHz mm^-2 is achieved. This is about an order of magnitude larger than the < 10 kHz mm^-2 of the nearest benchmark. Requiring no supplementary off-substrate signal-conditioning components and occupying an area of ≈11 mm² , the stand-alone AFE system is successfully applied to both electromyography and electrocardiography (ECG).

Musculoskeletal adaptation of young and older adults in response to challenging surface conditions

Over 36 million adults over 65 years of age experience accidental falls each year. The underlying neuromechanics (whole-body function) and driving forces behind accidental falls, as well as the effects of aging on the ability of the musculoskeletal system to adapt, are poorly understood. We evaluated differences in kinematics (lower extremity joint angles and range of motion), kinetics (ground reaction force), and electromyography (muscle co-contraction), due to changes in surface conditions during gait in 14 older adults with a history of falling and 14 young adults. We investigated the impact of challenging surfaces on musculoskeletal adaptation and compared the mechanisms of adaptation between age-groups. Older adults displayed greater hip and knee flexion and range of motion during gait, reduced initial vertical loading, and 13 % greater knee muscle co-contraction during early stance compared to young adults. Across age groups, the presence of an uneven challenging surface increased lower-limb flexion compared to an even surface. On a slick surface, older adults displayed 30 % greater ankle muscle co-contraction during early stance as compared to young adults. Older adults respond to challenging surfaces differently than their younger counterparts, employing greater flexion during early stance. This study underscores the need for determining lower-limb musculoskeletal adaptation strategies during gait and assessing how these strategies change with age, risk of accidental falls, and environmental surfaces to reduce the risk of accidental falls.

Do we all walk the walk? A comparison of walking behaviors across tetrapods

A walking gait has been identified in a range of vertebrate species with different body plans, habitats, and life histories. With increased application of this broad umbrella term, it has become necessary to assess the physical characteristics, analytical approaches, definitions, and diction used to describe walks. To do this, we reviewed studies of slow speed locomotion across a range of vertebrates to refine the parameters used to define walking, evaluate analytical techniques, and propose approaches to maximize consistency across subdisciplines. We summarize nine key parameters used to characterize walking behaviors in mammals, birds, reptiles, amphibians, and fishes. After identifying consistent patterns across groups, we propose a comprehensive definition for a walking gait. A walk is a form of locomotion where the majority of the forward propulsion of the animal comes from forces generated by the appendages interacting with the ground. During a walk, an appendage must be out of phase with the opposing limb in the same girdle and there is always at least one limb acting as ground-support (no suspension phase). Additionally, walking occurs at dimensionless speeds <1 v* and the duty factor of the limbs is always >0.5. Relative to other gaits used by the same species, the stance duration of a walk is long, the cycle frequency is low, and the cycle distance is small. Unfortunately, some of these biomechanical parameters, while effectively describing walks, may also characterize other, non-walking gaits. Inconsistent methodology likely contributes to difficulties in comparing data across many groups of animals; consistent application of data collection and analytical techniques in research methodology can improve these comparisons. Finally, we note that the kinetics of quadrupedal movements are still poorly understood and much work remains to be done to understand the movements of small, exothermic tetrapods.

A database of physical therapy exercises with variability of execution collected by wearable sensors

This document introduces the PHYTMO database, which contains data from physical therapies recorded with inertial sensors, including information from an optical reference system. PHYTMO includes the recording of 30 volunteers, aged between 20 and 70 years old. A total amount of 6 exercises and 3 gait variations were recorded. The volunteers performed two series with a minimum of 8 repetitions in each one. PHYTMO includes magneto-inertial data, together with a highly accurate location and orientation in the 3D space provided by the optical system. The files were stored in CSV format to ensure its usability. The aim of this dataset is the availability of data for two main purposes: the analysis of techniques for the identification and evaluation of exercises using inertial sensors and the validation of inertial sensor-based algorithms for human motion monitoring. Furthermore, the database stores enough data to apply Machine Learning-based algorithms. The participants' age range is large enough to establish age-based metrics for the exercises evaluation or the study of differences in motions between different groups.

Assessment of Stability of MIMU Probes to Skin-Marker-Based Anatomical Reference Frames During Locomotion Tasks: Effect of Different Locations on the Lower Limb

Soft tissue artefacts (STAs) undermine the validity of skin-mounted approaches to measure skeletal kinematics. Magneto-inertial measurement units (MIMU) gained popularity due to their low cost and ease of use. Although the reliability of different protocols for marker-based joint kinematics estimation has been widely reported, there are still no indications on where to place MIMU to minimize STA. This study aims to find the most stable positions for MIMU placement, among four positions on the thigh, four on the shank, and three on the foot. Stability was investigated by measuring MIMU movements against an anatomical reference frame, defined according to a standard marker-based approach. To this aim, markers were attached both on the case of each MIMU (technical frame) and on bony landmarks (anatomical frame). For each MIMU, the nine angles between each versor of the technical frame with each versor of the corresponding anatomical frame were computed. The maximum standard deviation of these angles was assumed as the instability index of MIMU-body coupling. Six healthy subjects were asked to perform barefoot gait, step negotiation, and sit-to-stand. Results showed that (1) in the thigh, the frontal position was the most stable in all tasks, especially in gait; (2) in the shank, the proximal position is the least stable, (3) lateral or medial calcaneus and foot dorsum positions showed equivalent stability performances. Further studies should be done before generalizing these conclusions to different motor tasks and MIMU-body fixation methods. The above results are of interest for both MIMU-based gait analysis and rehabilitation approaches using wearable sensors-based biofeedback.

A model-free deep reinforcement learning approach for control of exoskeleton gait patterns

Lower-body exoskeleton control that adapts to users and provides assistance-as-needed can increase user participation and motor learning and allow for more effective gait rehabilitation. Adaptive model-based control methods have previously been developed to consider a user’s interaction with an exoskeleton; however, the predefined dynamics models required are challenging to define accurately, due to the complex dynamics and nonlinearities of the human-exoskeleton interaction. Model-free deep reinforcement learning (DRL) approaches can provide accurate and robust control in robotics applications and have shown potential for lower-body exoskeletons. In this paper, we present a new model-free DRL method for end-to-end learning of desired gait patterns for over-ground gait rehabilitation with an exoskeleton. This control technique is the first to accurately track any gait pattern desired in physiotherapy without requiring a predefined dynamics model and is robust to varying post-stroke individuals’ baseline gait patterns and their interactions and perturbations. Simulated experiments of an exoskeleton paired to a musculoskeletal model show that the DRL method is robust to different post-stroke users and is able to accurately track desired gait pattern trajectories both seen and unseen in training.

Kinematic patterns during walking in children: Application of principal component analysis

The relative displacements of body segments during walking can be reduced to a small number of multi-joint kinematic patterns, pm_k, through Principal Component Analysis (PCA). These patterns were extracted from two groups of children (n = 8, aged 6-9 years, 4 males, and n = 8, aged 10-13 years, 4 males) and 7 adults (21-29 years, 1 male), walking on a treadmill at various velocities, normalized to body stature (adimensional Froude number, Fr). The three-dimensional coordinates of body markers were captured by an optoelectronic system. Five components (pm₁ to pm₅) explained 99.1% of the original dataset variance. The relationship between the variance explained ("size") of each pm_k and the Fr velocity varied across movement components and age groups. Only pm₁ and pm₂, which described kinematic patterns in the sagittal plane, showed significant differences (at p < 0.05) across pairs of age groups. The time course of the size of all the five components matched various mechanical events of the step cycle at the level of both body system and lower limb joints. Such movement components appeared clinically interpretable and lend themselves as potential markers of neural development of walking.

Lower-limb kinematics and kinetics during continuously varying human locomotion

Human locomotion involves continuously variable activities including walking, running, and stair climbing over a range of speeds and inclinations as well as sit-stand, walk-run, and walk-stairs transitions. Understanding the kinematics and kinetics of the lower limbs during continuously varying locomotion is fundamental to developing robotic prostheses and exoskeletons that assist in community ambulation. However, available datasets on human locomotion neglect transitions between activities and/or continuous variations in speed and inclination during these activities. This data paper reports a new dataset that includes the lower-limb kinematics and kinetics of ten able-bodied participants walking at multiple inclines (±0°; 5° and 10°) and speeds (0.8 m/s; 1 m/s; 1.2 m/s), running at multiple speeds (1.8 m/s; 2 m/s; 2.2 m/s and 2.4 m/s), walking and running with constant acceleration (±0.2; 0.5), and stair ascent/descent with multiple stair inclines (20°; 25°; 30° and 35°). This dataset also includes sit-stand transitions, walk-run transitions, and walk-stairs transitions. Data were recorded by a Vicon motion capture system and, for applicable tasks, a Bertec instrumented treadmill.

Kinematics, Speed, and Anthropometry-Based Ankle Joint Torque Estimation: A Deep Learning Regression Approach

Powered Assistive Devices (PADs) have been proposed to enable repetitive, user-oriented gait rehabilitation. They may include torque controllers that typically require reference joint torque trajectories to determine the most suitable level of assistance. However, a robust approach able to automatically estimate user-oriented reference joint torque trajectories, namely ankle torque, while considering the effects of varying walking speed, body mass, and height on the gait dynamics, is needed. This study evaluates the accuracy and generalization ability of two Deep Learning (DL) regressors (Long-Short Term Memory and Convolutional Neural Network (CNN)) to generate user-oriented reference ankle torque trajectories by innovatively customizing them according to the walking speed (ranging from 1.0 to 4.0 km/h) and users’ body height and mass (ranging from 1.51 to 1.83 m and 52.0 to 83.7 kg, respectively). Furthermore, this study hypothesizes that DL regressors can estimate joint torque without resourcing electromyography signals. CNN was the most robust algorithm (Normalized Root Mean Square Error: 0.70 ± 0.06; Spearman Correlation: 0.89 ± 0.03; Coefficient of Determination: 0.91 ± 0.03). No statistically significant differences were found in CNN accuracy (p-value > 0.05) whether electromyography signals are included as inputs or not, enabling a less obtrusive and accurate setup for torque estimation.

Lower limb kinematic, kinetic, and EMG data from young healthy humans during walking at controlled speeds

Understanding the lower limb kinematic, kinetic, and electromyography (EMG) data interrelation in controlled speeds is challenging for fully assessing human locomotion conditions. This paper provides a complete dataset with the above-mentioned raw and processed data simultaneously recorded for sixteen healthy participants walking on a 10 meter-flat surface at seven controlled speeds (1.0, 1.5, 2.0, 2.5, 3.0, 3.5, and 4.0 km/h). The raw data include 3D joint trajectories of 24 retro-reflective markers, ground reaction forces (GRF), force plate moments, center of pressures, and EMG signals from Tibialis Anterior, Gastrocnemius Lateralis, Biceps Femoris, and Vastus Lateralis. The processed data present gait cycle-normalized data including filtered EMG signals and their envelope, 3D GRF, joint angles, and torques. This study details the experimental setup and presents a brief validation of the data quality. The presented dataset may contribute to (i) validate and enhance human biomechanical gait models, and (ii) serve as a reference trajectory for personalized control of robotic assistive devices, aiming an adequate assistance level adjusted to the gait speed and user's anthropometry.

A comprehensive, open-source dataset of lower limb biomechanics in multiple conditions of stairs, ramps, and level-ground ambulation and transitions

We introduce a novel dataset containing 3-dimensional biomechanical and wearable sensor data from 22 able-bodied adults for multiple locomotion modes (level-ground/treadmill walking, stair ascent/descent, and ramp ascent/descent) and multiple terrain conditions of each mode (walking speed, stair height, and ramp inclination). In this paper, we present the data collection methods, explain the structure of the open dataset, and report the sensor data along with the kinematic and kinetic profiles of joint biomechanics as a function of the gait phase. This dataset offers a comprehensive source of locomotion information for the same set of subjects to motivate applications in locomotion recognition, developments in robotic assistive devices, and improvement of biomimetic controllers that better adapt to terrain conditions. With such a dataset, models for these applications can be either subject-dependent or subject-independent, allowing greater flexibility for researchers to advance the field.

SENSITIVITY OF MODEL-PREDICTED MUSCLE FORCES OF PATIENTS WITH CEREBRAL PALSY TO VARIATIONS IN MUSCLE-TENDON PARAMETERS

Computational musculoskeletal modeling and simulation platforms are efficient tools to gain insight into the muscular coordination of patients with motor disabilities such as cerebral palsy (CP). Muscle force predictions from simulation programs are influenced by the architectural and contractile properties of muscle-tendon units. In this study, we aimed to evaluate the sensitivity of major lower limb muscle forces in patients with CP to changes in muscle-tendon parameters. Open-access datasets of children with CP ([Formula: see text]) and healthy children ([Formula: see text]) were considered. Monte Carlo analysis was executed to specify how sensitive the muscle forces to perturbations between [Formula: see text]% and [Formula: see text]% of the nominal value of the maximum isometric muscle force, optimal muscle fiber length, muscle pennation angle, tendon slack length, and maximum contraction velocity of muscle. The sensitivity analysis revealed that muscle forces of CP patients and healthy individuals were most sensitive to perturbations in the tendon slack length ([Formula: see text]), while forces of CP patients were more sensitive to tendon slack length when compared to the healthy group ([Formula: see text]). Muscle forces of patients and healthy individuals were insensitive to the other four parameters ([Formula: see text]), except for the gracilis and sartorius muscles in which the proportion of optimal muscle fiber length to tendon slack length is higher than 1; forces of these two muscles were also sensitive to the optimal muscle fiber length. The results of this study are expected to contribute to our understanding of which parameters should be personalized when conducting musculoskeletal modeling and simulation of patients with CP.

Symmetry function in gait pattern analysis in patients after unilateral transfemoral amputation using a mechanical or microprocessor prosthetic knee

BACKGROUND

Above-knee amputations (AKAs) contribute to gait asymmetry. The level of asymmetry is affected by the type of knee prosthetic module. There is limited evidence suggesting that more technically advanced solutions (microprocessor modules; MicPK) are superior to less advanced ones (mechanical modules; MechPK). The study aimed to evaluate the variable range of hip and pelvic joint movements during gait and look for differentiating areas with an increased level of asymmetry of the gait cycle in individuals who underwent an AKA and are equipped with MicPK or MechPK.

METHODS

Twenty-eight individuals participated in the study; 14 were assigned to a study group of individuals who underwent a unilateral AKA, and the other 14 were healthy participants as a control group. The movement task was recorded using the optoelectronic SMART-E system following the standard Davis protocol (the Newington model). A new method of quantifying gait symmetry using the symmetry function (SF) is proposed. SF is an integral measure of absolute differences in time-standardized signals between sides throughout the whole cycle of motion variability.

RESULTS

In the frontal plane, there were significant differences between groups in the asymmetry of the range of movement in the hip joint of the intact limb. In the middle of the support phase, the intact limb was more adducted in individuals with MicPK and less abducted in people with MechPK (differences in mean 180%, p < 0.000; max 63%, p < 0.000; min 65%, p < 0.000). In the sagittal plane, the range of asymmetry of the flexion and thigh extension of the intact limb was similar to and only slightly different from the physiological gait. In the transverse plane, higher asymmetry values were noted for individuals with MicPK. In the final stage of the swing phase, the thigh was more rotated both externally and internally. The size of the asymmetry, when compared to gait of healthy individuals, reached 50% (differences in mean 115%, p < 0.232; max 62% p < 0.26; min 50, p < 0.154).

CONCLUSIONS

In the study group, the assessed ranges of pelvic and thigh movement in the hip joint differed only in the frontal plane. Individuals who underwent a unilateral above-knee amputation may gain less from using MicPK than anticipated.

Optical motion capture dataset of selected techniques in beginner and advanced Kyokushin karate athletes

Human motion capture is commonly used in various fields, including sport, to analyze, understand, and synthesize kinematic and kinetic data. Specialized computer vision and marker-based optical motion capture techniques constitute the gold-standard for accurate and robust human motion capture. The dataset presented consists of recordings of 37 Kyokushin karate athletes of different ages (children, young people, and adults) and skill levels (from 4th dan to 9th kyu) executing the following techniques: reverse lunge punch (Gyaku-Zuki), front kick (Mae-Geri), roundhouse kick (Mawashi-Geri), and spinning back kick (Ushiro-Mawashi-Geri). Each technique was performed approximately three times per recording (i.e., to create a single data file), and under three conditions where participants kicked or punched (i) in the air, (ii) a training shield, or (iii) an opponent. Each participant undertook a minimum of two trials per condition. The data presented was captured using a Vicon optical motion capture system with Plug-In Gait software. Three dimensional trajectories of 39 reflective markers were recorded. The resultant dataset contains a total of 1,411 recordings, with 3,229 single kicks and punches. The recordings are available in C3D file format. The dataset provides the opportunity for kinematic analysis of different combat sport techniques in attacking and defensive situations.

Development of a neural network based control algorithm for powered ankle prosthesis

Lower limb amputation is partial or complete removal of the limb due to disease, accident or trauma. Surface electromyograms (sEMG) of a large number of muscles and force sensors have been used to develop control algorithms for lower limb powered prostheses, but there are no commercial sEMG controlled prostheses available to date. Unlike ankle disarticulation, transtibial amputation yields less intact lower leg muscle mass. Therefore, minimizing the use of sEMG muscle sources utilized will make powered prosthesis controller economic, and limiting the use of specifically the lower leg muscles will make it flexible. Presently, we have used healthy population data to (1) test the feasibility of the neural network (NN) approach for developing a powered ankle prosthesis control algorithm that successfully predicts sagittal ankle angle and moment during walking using exclusively sEMG, and (2) rank all muscle combination variations according to their success to determine the economic and flexible NN's. sEMG amplitudes of five lower extremity muscles were used as inputs: the tibialis anterior (TA), medial gastrocnemius (MG), rectus femoris (RF), biceps femoris (BF) and gluteus maximus (GM). A time-delay feed-forward-multilayer-architecture NN algorithm was developed. Muscle combination variations were ranked using Pearson's correlation coefficient (r > 0.95 indicates successful correlations) and root-mean-square error between actual vs. estimated ankle position and moment. The trained NN TA + MG was successful (r_position = 0.952, r_moment = 0.997) whereas, TA + MG + BF (r_position = 0.981, r_moment = 0.996) and MG + BF + GM (r_position = 0.955, r_moment = 0.995) were distinguished as the economic and flexible variations, respectively. The algorithms developed should be trained and tested for data acquired from amputees in new studies.

Elbow movement estimation based on EMG with NARX Neural Networks

The use of the electrical activity from the muscles may provide a natural way to control exoskeletons or other robotic devices seamlessly. The major challenges to achieve this goal are human motor redundancy and surface electromyography (sEMG) variability. The goal of this work is to find a feature extraction and classification procedures to estimate accurately elbow angular trajectory by means of a NARX Neural Network. The processing time-step should be small enough to make it feasible its further use for online control of an exoskeleton. In order to do so we analysed the Biceps and Triceps Brachii data from an elbow flexo-extension Coincident Timing task performed in the horizontal plane. The sEMG data was pre-processed and its energy was divided in five frequency intervals that were fed to a Nonlinear Auto Regressive with Exogenous inputs (NARX) Neural Network. The estimated angular trajectory was compared with the measured one showing a high correlation between them and a RMSE error maximum of 7 degrees. The procedure presented here shows a reasonably good estimation that, after training, allows real-time implementation. In addition, the results are encouraging to include more complex tasks including the shoulder joint.

Robotic Rehabilitation in Spinal Cord Injury: A Pilot Study on End-Effectors and Neurophysiological Outcomes

Robot-aided gait training (RAGT) has been implemented to provide patients with spinal cord injury (SCI) with a physiological limb activation during gait, cognitive engagement, and an appropriate stimulation of peripheral receptors, which are essential to entrain neuroplasticity mechanisms supporting functional recovery. We aimed at assessing whether RAGT by means of an end-effector device equipped with body weight support could improve functional ambulation in patients with subacute, motor incomplete SCI. In this pilot study, 15 patients were provided with six RAGT sessions per week for eight consecutive weeks. The outcome measures were muscle strength, ambulation, going upstairs, and disease burden. Furthermore, we estimated the activation patterns of lower limb muscles during RAGT by means of surface electromyography and the resting state networks' functional connectivity (RSN-FC) before and after RAGT. Patients achieved a clinically significant improvement in the clinical outcome measures substantially up to six months post-treatment. These data were paralleled by an improvement in the stair-climbing cycle and a potentiating of frequency-specific and area-specific RSN-FC patterns. Therefore, RAGT, by means of an end-effector device equipped with body weight support, is promising in improving gait in patients with subacute, motor incomplete SCI, and it could produce additive benefit for the neuromuscular reeducation to gait in SCI when combined with conventional physiotherapy.

Improved Gait of Persons With Multiple Sclerosis After Rehabilitation: Effects on Lower Limb Muscle Synergies, Push-Off, and Toe-Clearance

Introduction: Persons with MS (PwMS) have markedly reduced push-off and toe-clearance during gait compared to healthy subjects (HS). These deficits may result from alterations in neuromotor control at the ankle. To optimize rehabilitation interventions for PwMS, a crucial step is to evaluate if and how altered neuromotor control, as represented by muscle synergies, improves with rehabilitation. In this study we investigated changes in ankle motor control and associated biomechanical parameters during gait in PwMS, occurring with increase in speed after gait rehabilitation. Methods: 3D motion and EMG data were collected while 11 PwMS (age 50.3 + 11.1; EDSS 5.2 + 1.2) walked overground at self-selected speed before (T0) and after 20 sessions (T1) of intensive treadmill training. Muscle synergies were extracted using non-negative matrix factorization. Gait parameters were computed according to the LAMB protocol. Pearson's correlation coefficient was used to evaluate the similarity of motor modules between PwMS and HS. To assess differences in distal module activations representing neuromotor control at the ankle [Forward Propulsion (FPM) and Ground Clearance modules (GCM)], each module's activation timing was integrated over 100% of the gait cycle and the activation percentage index (API) was computed in six phases. Ten age matched HS provided two separate speed-matched normative datasets for T0 and T1. For speed independent comparison for the PwMs Z scores were calculated for all their gait variables. Results: In PwMS velocity increased significantly from T0 to T1 (0.74-0.90 m/s, p < 0.05). The activation profiles (API) of FPM and GCM of PwMS improved in pre-swing (p < 0.05): FPM (Mean [95% CI] [%]: T0: 12.5 [5.7-19.3] vs. T1: 9.0 [2.7-15.3]); GCM (T0: 26.7 [18.2-35.3] vs. T1: 24.5 [18.2-30.7]). This was associated with an increase in toe clearance (80.3 to 103.6 mm, p < 0.05) and a higher ankle power peak in pre-swing (1.53-1.93 W/kg, p < 0.05). Conclusion: Increased gait speed of PwMS after intensive gait training was consistent with improvements in spatio-temporal gait parameters. The most important finding of this study was the re-organization of distal leg modules related to neurophysiological changes induced by rehabilitation. This was associated with an improved ankle performance.

Wireless Epidermal Electromyogram Sensing System

Massive efforts to build walking aid platforms for the disabled have been made in line with the needs of the aging society. One of the core technologies that make up these platforms is a realization of the skin-like electronic patch, which is capable of sensing electromyogram (EMG) and delivering feedback information to the soft, lightweight, and wearable exosuits, while maintaining high signal-to-noise ratio reliably in the long term. The main limitations of the conventional EMG sensing platforms include the need to apply foam tape or conductive gel on the surface of the device for adhesion and signal acquisition, and also the bulky size and weight of conventional measuring instruments for EMG, limiting practical use in daily life. Herein, we developed an epidermal EMG electrode integrated with a wireless measuring system. Such the stretchable platform was realized by transfer-printing of the as-prepared EMG electrodes on a SiO2 wafer to a polydimethylsiloxane (PDMS) elastomer substrate. The epidermal EMG patch has skin-like properties owing to its unique mechanical characteristics: i) location on a neutral mechanical plane that enables high flexibility, ii) wavy design that allows for high stretchability. We demonstrated wireless EMG monitoring using our skin-attachable and stretchable EMG patch sensor integrated with the miniaturized wireless system modules.