Human quiet stance appears to be a simple postural task, yet it relies on complex neural control mechanisms to maintain balance. Although quiet stance control has been studied extensively, the underlying stabilization strategy remains debated. Intermittent neural feedback control is one of the few plausible mechanisms that can simultaneously stabilize upright posture and account for the characteristic sway observed during standing. However, intermittent control during quiet stance constitutes a distinctive class of nonlinear control characterized by state-dependent switching and manifold-triggered activation. These properties complicate the application of conventional nonlinear control frameworks to rigorous stability analysis. Consequently, much of the existing literature has relied predominantly on numerical simulations rather than analytical results. In this study, we systematically investigated the stability of quiet stance under manifold-triggered intermittent neural feedback control, with particular emphasis on marginal stability. We found that the control manifold destabilized the system in some activation regions but enhanced stability in others. To address the theoretical challenges posed by delays and stochastic noise, we further developed a statistical-physics-based approach for stability characterization that complements Lyapunov-based analyses and is applicable to noise-perturbed delayed dynamics. Using this framework, we delineated the stable parameter regions for intermittent neural control and quantified how key system parameters shape the stability boundaries through both theoretical analysis and numerical validation. Together, these results provide a principled perspective for studying noise-perturbed delayed nonlinear dynamics in intermittent control models of quiet stance, offering insights into neural balance regulation and informing potential applications such as fall-risk mitigation and the control design of lower-limb exoskeletons.

Human-machine interfaces (HMIs) have been widely integrated with motor rehabilitation and augmentation systems. Forecasting movement transitions during human-robot interaction is crucial to ensure system safety, intuitiveness, and reactivity, particularly in anticipating human motor intentions under sudden perturbations or emergency scenarios. In this study, we investigated pre-movement neural signatures preceding sudden movement transitions during ongoing bimanual tasks. Informed by these findings, we propose a physiology-informed EEG Transformer (PI-EEGformer) for EEG-based motor intention recognition. An EEG dataset collected from a bimanual movement task, where one hand was required to switch motor states in response to unexpected cues, was used to evaluate the performance of the PI-EEGformer in comparison with seven state-of-the-art models. Results showed that, prior to the movement transition, EEG power spectrum decreased, and movement-related cortical potentials (MRCPs) could be accurately extracted from the contralateral motor cortex. PI-EEGformer reached an average accuracy of 0.912 in inter-subject tests and 0.829 in cross-subject tests in detecting movement transitions using EEG from 500 ms to 100 ms prior to the actual movement. This performance was superior to all the state-of-the-art models tested. These results demonstrate that EEG neural signatures can predict sudden movement transitions during ongoing bimanual tasks. The PI-EEGformer, designed with these physiological signatures, can enable accurate prediction of sudden movement transitions. This study will help improve the response of HMI systems to sudden disturbances, contributing to a more realistic HMI system.

Neural source separation enables the extraction of individual spike trains from complex electrophysiological recordings. When applied to electromyographic (EMG) signals, it provides a unique window into the motor output of the nervous system by isolating the spiking activity of motor units (MUs). MU decomposition from EMG signals is currently the only scalable neural interfacing approach available in behaving humans and has become foundational in motor neuroscience and neuroprosthetics. However, unlike related domains such as spike sorting or electroencephalography (EEG) analysis, decomposition of EMG signals lacks open benchmarks that reflect the diversity of muscles, movement contexts, and noise sources encountered in practice.To address this gap, we introduce MUniverse, a modular simulation and benchmarking suite for decomposing EMG signals into individual MU spiking activity. MUniverse provides: (1) a simulation stack with a user-friendly interface to a state-of-the-art EMG generator; (2) a curated library of datasets across synthetic, hybrid synthetic-real data with ground truth spikes, and experimental EMG; (3) a set of internal and external decomposition pipelines; and (4) a unified benchmark with well-defined tasks, standard evaluation metrics, and baseline results from established decomposition pipelines.MUniverse is designed for extensibility, reproducibility, and community use, and all datasets are distributed with standardised metadata (Croissant, BIDS). By standardising evaluation and enabling dataset simulation at scale, MUniverse aims to catalyze progress on this long-standing neural signal processing problem.

Both closed-loop models and multi-input multi-output (MIMO) models of the neuromusculoskeletal system of the upper limb are important for simulating and understanding motor control. Yet no large-scale linear neuromusculoskeletal models of the upper limb that are both closed-loop and MIMO have been developed. The primary difficulty in creating such models is choosing appropriate feedback parameters (such as feedback gains and delays), as such a collection of parameters is not available in the literature. The purpose of this work is to 1) present a method for developing MIMO models of short-loop afferent feedback and 2) offer estimates of average feedback parameter values and ranges based on the currently available literature. To this end, we combined measurements of feedback-related parameters available in 26 prior studies with known properties of system stability and behavior. As a result, we present estimated feedback gains and delays for a linear model of the upper limb with inputs into the 13 major superficial muscles and outputs to the 7 main joint degrees of freedom from the shoulder to the wrist. This model includes homonymous feedback mediated by Golgi tendon organs and both homonymous and heteronymous feedback mediated by muscle spindles. As a partial validation of muscle-spindle feedback gains, we compared the sign of the estimated gains to known differences in excess central delay between excitatory and inhibitory connections. The comparison proved correct in all 39 muscle pairs for which we had both estimated a feedback gain and found a measured excess central delay value in the literature. Furthermore, as a partial validation of delay times, we compared estimated delay times to measured innervation lengths. We found a strong fit for efferent delays (R = 0.88) and a moderate fit for afferent delays (R = 0.65). In addition, we demonstrate the effect of feedback on model behavior and present brief comparisons between this behavior and experimentally observed behaviors of the human upper limb with and without feedback.

As a primary effector of humans, the hand plays a crucial role in many aspects of daily life. Recognizing multidegree-of-freedom hand movements from muscle activity helps infer human motion intentions. Solving this problem has direct applications in prosthetic and exoskeleton control. Here, we propose a self-supervised learning algorithm inspired by muscle synergies to achieve simultaneous estimation of wrist rotation (supination/pronation) and hand grasp (open/close) from sonomyography-the muscle deformation detected by a wearable ultrasound array. Unlike conventional methods collecting both muscle activity and hand kinematics for supervised model calibration, this algorithm only uses unlabeled forearm ultrasound signals for self-supervised wrist and hand movement estimation, where movement labels are auto-generated. The performance of the proposed algorithm was experimentally evaluated with ten participants including an amputee. Offline analysis demonstrated that the proposed algorithm can accurately estimate simultaneous wrist rotation and hand grasp movements and were 0.98 and 0.94 for the able-bodied, and 0.98 and 0.90 for the amputee, respectively). Notably, the performance of the self-supervised learning was superior to the supervised learning for the amputee. Online experiments demonstrated that intended wrist and hand movements can be deciphered in real time, enabling accurate control of a virtual hand. This study will open up a new avenue for the sonomyographic human-machine interaction.

Prediction of transition between locomotion modes (e.g. moving from flat ground to stairs, etc) is vital for optimal interface with lower limb assistive technologies such as exoskeletons and prostheses. Inertial and bipolar electromyography (EMG) sensors have been investigated, but accuracy for clinical utility remains unresolved. This shortfall may be attributed to their limited capacity to detect subtle changes in muscle activations, particularly during the early stages of locomotion transitions (e.g., near the toe-off). In this study, we examined the effectiveness of two high-density surface electromyography (HDsEMG) sensors in detecting muscle activation changes during stair-related transitions. The results revealed that compared to bipolar EMG on the same muscles, HDsEMG-based methods increased transition prediction accuracy significantly from 70.2% to 91.1% when predicting at toe-off and from 89.8% to 99.2% when predicting with a delay of 400-ms relative to toe-off. This demonstrated the superior ability of HDsEMG to capture subtle muscle activation changes, especially during early transition stages. We also found reducing the electrode count to 21 per muscle only minimally impacted performance (88.3% accuracy at toe-off). This suggests distributing the same total number of electrodes across more muscles could potentially further improve prediction accuracy without increasing computational load. Moreover, by implementing image-inpainting signal processing, HDsEMG demonstrated robustness against the common issue of electrode signal loss. Even with 30% electrode detachment, prediction accuracy decreased only by 3%. We argue that HDsEMG offers a promising solution to bridge the gap in locomotion transition prediction for interface with assistive technology.

Hand paralysis due to spinal cord injury (SCI) greatly limits the quality of life of injured individuals. Despite complete loss of hand digit control, however, residual electrical muscle activity is often detected from these injured individuals. From this activity, individual motor unit action potentials can be identified and potentially used to infer their motion intent for interfacing purposes. We recently demonstrated that residual motor units can be decoded from tetraplegic individuals with SCI, by mapping both proximal and distal forearm activity using hundreds of electromyography (EMG) electrodes. Yet, few explored the feasibility of neural interfacing using only forearm motor units or even far-field wrist motor units in SCI, which will facilitate the use of fully wearable systems such as EMG bracelets. Here, we recognize finger gestures in eight tetraplegic individuals (Seven with motor complete SCI and one with motor incomplete SCI), using either forearm or wrist motor units. We demonstrate that motion-wise surface EMG decomposition can effectively increase the number of decomposed motor units from both forearm and wrist (on average 41.25 $\pm$ 24.14 from the forearm and 30 $\pm$ 9.72 from the wrist) and to reach high accuracy in gesture recognition at both locations (82% to 100% with the forearm data, and 62% to 99 with the wrist data). The decomposition met the requirement of real-time implementation. Moreover, the correlation between far-field motor units activity recorded from the wrist with the activity recorded at the forearm is revealed, further suggesting both locations are suitable for interfacing.

Numerical models of electromyography (EMG) signals have provided a huge contribution to our fundamental understanding of human neurophysiology and remain a central pillar of motor neuroscience and the development of human-machine interfaces. However, while modern biophysical simulations based on finite element methods (FEMs) are highly accurate, they are extremely computationally expensive and thus are generally limited to modeling static systems such as isometrically contracting limbs. As a solution to this problem, we propose to use a conditional generative model to mimic the output of an advanced numerical model. To this end, we present BioMime, a conditional generative neural network trained adversarially to generate motor unit (MU) activation potential waveforms under a wide variety of volume conductor parameters. We demonstrate the ability of such a model to predictively interpolate between a much smaller number of numerical model's outputs with a high accuracy. Consequently, the computational load is dramatically reduced, which allows the rapid simulation of EMG signals during truly dynamic and naturalistic movements.

sEMG pattern recognition algorithms have been explored extensively in decoding movement intent, yet are known to be vulnerable to changing recording conditions, exhibiting significant drops in performance across subjects, and even across sessions. Multi-channel surface EMG, also referred to as high-density sEMG (HD-sEMG) systems, have been used to improve performance with the information collected through the use of additional electrodes. However, a lack of robustness is ever present due to limited datasets and the difficulties in addressing sources of variability, such as electrode placement. In this study, we propose training on a collection of input channel subsets and augmenting our training distribution with data from different electrode locations, simultaneously targeting electrode shift and reducing input dimensionality. Our method increases robustness against electrode shift and results in significantly higher intersession performance across subjects and classification algorithms.

This work presents a wearable, eight-channel A-mode ultrasound system for hand gesture recognition and interactive gaming. The wearable system consists of a custom-built forearm bracelet with eight piezoelectric transducers (1 MHz) evenly distributed, an electronic hub housed on a bicep strap, a laptop with a bespoke machine learning algorithm and a robotic hand for actuation. The electronic hub drives the transducers to produce ultrasound pulses that will travel into the forearm muscles. The reflected ultrasound signals will vary based on the forearm muscular morphology (movement-dependent). By decoding the reflected ultrasound signals using machine learning techniques, a robotic hand can be controlled based on the user’s hand motions. In contrast to the conventional surface electromyography methods, the visitor can experience seamless control of a robotic hand and a maze navigation game via ultrasound as a novel sensing modality.

The cortico-spinal neural pathway is fundamental for motor control and movement execution, and in humans it is typically studied using concurrent electroencephalography (EEG) and electromyography (EMG) recordings. However, current approaches for capturing high-level and contextual connectivity between these recordings have important limitations. Here, we present a novel application of statistical dependence estimators based on orthonormal decomposition of density ratios to model the relationship between cortical and muscle oscillations. Our method extends from traditional scalar-valued measures by learning eigenvalues, eigenfunctions, and projection spaces of density ratios from realizations of the signal, addressing the interpretability, scalability, and local temporal dependence of cortico-muscular connectivity. We experimentally demonstrate that eigenfunctions learned from cortico-muscular connectivity can accurately classify movements and subjects. Moreover, they reveal channel and temporal dependencies that confirm the activation of specific EEG channels during movement.

Neuromechanical studies investigate how the nervous system interacts with the musculoskeletal (MSK) system to generate volitional movements. Such studies have been supported by simulation models that provide insights into variables that cannot be measured experimentally and allow a large number of conditions to be tested before the experimental analysis. However, current simulation models of electromyography (EMG), a core physiological signal in neuromechanical analyses, remain either limited in accuracy and conditions or are computationally heavy to apply. Here, we provide a computational platform to enable future work to overcome these limitations by presenting NeuroMotion, an open-source simulator that can modularly test a variety of approaches to the full-spectrum synthesis of EMG signals during voluntary movements. We demonstrate NeuroMotion using three sample modules. The first module is an upper-limb MSK model with OpenSim API to estimate the muscle fibre lengths and muscle activations during movements. The second module is BioMime, a deep neural network-based EMG generator that receives nonstationary physiological parameter inputs, like the afore-estimated muscle fibre lengths, and efficiently outputs motor unit action potentials (MUAPs). The third module is a motor unit pool model that transforms the muscle activations into discharge timings of motor units. The discharge timings are convolved with the output of BioMime to simulate EMG signals during the movement. We first show how MUAP waveforms change during different levels of physiological parameter variations and different movements. We then show that the synthetic EMG signals during two-degree-of-freedom hand and wrist movements can be used to augment experimental data for regressing joint angles. Ridge regressors trained on the synthetic dataset were directly used to predict joint angles from experimental data. In this way, NeuroMotion was able to generate full-spectrum EMG for the first use-case of human forearm electrophysiology during voluntary hand, wrist, and forearm movements. All intermediate variables are available, which allows the user to study cause-effect relationships in the complex neuromechanical system, fast iterate algorithms before collecting experimental data, and validate algorithms that estimate non-measurable parameters in experiments. We expect this modular platform will enable validation of generative EMG models, complement experimental approaches and empower neuromechanical research.

Automated source separation algorithms have become a central tool in neuroengineering and neuroscience, where they are used to decompose neurophysiological signal into its constituent spiking sources. However, in noisy or highly multivariate recordings these decomposition techniques often make a large number of errors. Such mistakes degrade online human-machine interfacing methods and require costly post-hoc manual cleaning in the offline setting. In this article we propose an automated error correction methodology using a deep metric learning (DML) framework, generating embedding spaces in which spiking events can be both identified and assigned to their respective sources. Furthermore, we investigate the relative ability of different DML techniques to preserve the intraclass semantic structure needed to identify incorrect class labels in neurophysiological time series. Motivated by this analysis, we propose locality sensitive mining, an easily implemented sampling-based augmentation to typical DML losses which substantially improves the local semantic structure of the embedding space. We demonstrate the utility of this method to generate embedding spaces which can be used to automatically identify incorrectly labeled spiking events with high accuracy.

To control wearable robotic systems, it is critical to obtain a prediction of the user's motion intent with high accuracy. Surface electromyography (sEMG) recordings have often been used as inputs for these devices, however bipolar sEMG electrodes are highly sensitive to their location. Positional shifts of electrodes after training gait prediction models can therefore result in severe performance degradation. This study uses high-density sEMG (HD-sEMG) electrodes to simulate various bipolar electrode signals from four leg muscles during steady-state walking. The bipolar signals were ranked based on the consistency of the corresponding sEMG envelope's activity and timing across gait cycles. The locations were then compared by evaluating the performance of an offline temporal convolutional network (TCN) that mapped sEMG signals to knee angles. The results showed that electrode locations with consistent sEMG envelopes resulted in greater prediction accuracy compared to hand-aligned placements (p$< $0.01). However, performance gains through this process were limited, and did not resolve the position shift issue. Instead of training a model for a single location, we showed that randomly sampling bipolar combinations across the HD-sEMG grid during training mitigated this effect. Models trained with this method generalized over all positions, and achieved 70% less prediction error than location specific models over the entire area of the grid. Therefore, the use of HD-sEMG grids to build training datasets could enable the development of models robust to spatial variations, and reduce the impact of muscle-specific electrode placement on accuracy.

Brain-computer interfaces (BCIs) provide a direct pathway from the brain to external devices and have demonstrated great potential for assistive and rehabilitation technologies. Endogenous BCIs based on electroencephalogram (EEG) signals, such as motor imagery (MI) BCIs, can provide some level of control. However, mastering spontaneous BCI control requires the users to generate discriminative and stable brain signal patterns by imagery, which is challenging and is usually achieved over a very long training time (weeks/months). Here, we propose a human-machine joint learning framework to boost the learning process in endogenous BCIs, by guiding the user to generate brain signals toward an optimal distribution estimated by the decoder, given the historical brain signals of the user. To this end, we first model the human-machine joint learning process in a uniform formulation. Then a human-machine joint learning framework is proposed: 1) for the human side, we model the learning process in a sequential trial-and-error scenario and propose a novel "copy/new" feedback paradigm to help shape the signal generation of the subject toward the optimal distribution and 2) for the machine side, we propose a novel adaptive learning algorithm to learn an optimal signal distribution along with the subject's learning process. Specifically, the decoder reweighs the brain signals generated by the subject to focus more on "good" samples to cope with the learning process of the subject. Online and psuedo-online BCI experiments with 18 healthy subjects demonstrated the advantages of the proposed joint learning process over coadaptive approaches in both learning efficiency and effectiveness.

Neurorobotic devices, such as prostheses, exoskeletons, and muscle stimulators, can partly restore motor functions in individuals with disabilities, such as stroke, spinal cord injury (SCI), and amputations and musculoskeletal impairments. These devices require information transfer from and to the nervous system by neurorobotic interfaces. However, current interfacing systems have limitations of low-spatial and temporal resolution, and lack robustness, with sensitivity to, e.g., fatigue and sensor displacement. Muscle scanning and imaging by ultrasound technology has emerged as a neurorobotic interface alternative to more conventional electrophysiological recordings. While muscle ultrasound detects movement of muscle fibers, and therefore does not directly detect neural information, the muscle fibers are activated by neurons in the spinal cord and therefore their motions mirror the neural code sent from the spinal cord to muscles. In this view, muscle imaging by ultrasound provides information on the neural activation underlying movement intent and execution. Here, we critically review the literature on ultrasound applied as a neurorobotic interface, focusing on technological progresses and current achievements, machine learning algorithms, and applications in both upper-and lower-limb robotics. This critical review reveals that ultrasound in the human-machine interface field has evolved from bulky hardware to miniaturized systems, from multichannel imaging to sparse channel sensing, from simple muscle morphological analysis to input signal for musculoskeletal models and machine learning, from unimodal sensing to multimodal fusion, and from conventional statistical learning to deep learning. For future advances, we recommend exploring high-precision ultrasound imaging technology, improving the wearability and ergonomics of systems and transducers, and developing user-friendly real-time human-machine interaction models.

The computational simulation of human voluntary muscle contraction is possible with EMG-driven Hill-type models of whole muscles. Despite impactful applications in numerous fields, the neuromechanical information and the physiological accuracy such models provide remain limited because of multiscale simplifications that limit comprehensive description of muscle internal dynamics during contraction. We addressed this limitation by developing a novel motoneuron-driven neuromuscular model, that describes the force-generating dynamics of a population of individual motor units, each of which was described with a Hill-type actuator and controlled by a dedicated experimentally derived motoneuronal control. In forward simulation of human voluntary muscle contraction, the model transforms a vector of motoneuron spike trains decoded from high-density EMG signals into a vector of motor unit forces that sum into the predicted whole muscle force. The motoneuronal control provides comprehensive and separate descriptions of the dynamics of motor unit recruitment and discharge and decodes the subject's intention. The neuromuscular model is subject-specific, muscle-specific, includes an advanced and physiological description of motor unit activation dynamics, and is validated against an experimental muscle force. Accurate force predictions were obtained when the vector of experimental neural controls was representative of the discharge activity of the complete motor unit pool. This was achieved with large and dense grids of EMG electrodes during medium-force contractions or with computational methods that physiologically estimate the discharge activity of the motor units that were not identified experimentally. This neuromuscular model advances the state-of-the-art of neuromuscular modelling, bringing together the fields of motor control and musculoskeletal modelling, and finding applications in neuromuscular control and human-machine interfacing research.

Simultaneous and proportional control (SPC) based on surface electromyographic (sEMG) signals has led to a broad range of applications. However, due to the limitation in the generalization and stability of current machine learning algorithms, these methods can only estimate less than 15 simultaneuous and proportional (SP) categories of finger movement. In this article, a novel deep learning algorithm, named multiattention feature fusion network (MAFN), is proposed to estimate comprehensive finger movement (up to 28 categories SP movements) from sEMG signals. MAFN is based on the multihead attention mechanism, which adaptively extracts essential features for analyzing the joint angles from the extracted sEMG features. Furthermore, a real-time exponential smoothing algorithm is designed for further improvement of the prediction stability. MAFN was evaluated on 28 finger movements of 38 subjects in the Ninapro_db2 dataset, and benchmarked with the state-of-the-art methods, such as temporal convolutional network (TCN) and long short term memory network (LSTM). The results demonstrated that the average Pearson correlation coefficient, root mean squared error of MAFN (0.84 ± 0.03,0.09 ± 0.01) were significantly higher than those of TCN (0.52 ± 0.06,pppp< 0.001). These improvements led to more stable and accurate movement predictions. Additionally, the time delay and power consumption of MAFN when applied to sEMG signals on a portable device are only 83.4 ms and 3 W, which implies prospective commercial applications.

We investigate the use of high-density surface EMG (HDsEMG) recordings of intrinsic hand muscles, along with those from extrinsic muscles, on finger and wrist kinematic prediction performance. We incorporate these HDsEMG signals using a framework based on a custom hybrid convolutional-recurrent deep learning model. Methods: Five healthy subjects performed a wide variety of motion tasks activating multiple degrees of freedom of the wrist and fingers. During the tasks, HDsEMG signals were recorded from extrinsic and intrinsic muscles of the hand while motion capture technology tracked the hand/wrist kinematics. A convolutional-recurrent model architecture was designed and trained on the recorded dataset, incorporating both residual connections as well as inception convolutional structures. Results: The proposed model led to greater regression accuracy over the simultaneous prediction of 12 joint angles (CC, MAE and RMSE of 0.850, 4.84 degrees and 11.2 degrees respectively) than previously proposed mapping models, when incorporating both intrinsic and extrinsic muscle signals. The inclusion of both sets of hand muscles also led to statistically greater performance than the same model trained on only extrinsic muscle data. Conclusion: We show accurate predictions of hand/wrist kinematics from combined extrinsic and intrinsic hand muscle myoelectric activity, using a convolutionalrecurrent hybrid deep learning model. This greater performance is replicated over several subjects and across multidegree of freedom motion tasks. Significance: Our developed system (electrode setup and deep neural networks) can be translated into a compact wearable interface in the future for medical as well as consumer applications.

Advances in biosignal signal processing and machine learning, in particular Deep Neural Networks (DNNs), have paved the way for the development of innovative Human-Machine Interfaces for decoding the human intent and controlling artificial limbs. DNN models have shown promising results with respect to other algorithms for decoding muscle electrical activity, especially for recognition of hand gestures. Such data-driven models, however, have been challenged by their need for a large number of trainable parameters and their structural complexity. Here we propose the novel Temporal Convolutions-based Hand Gesture Recognition architecture (TC-HGR) to reduce this computational burden. With this approach, we classified 17 hand gestures via surface Electromyogram (sEMG) signals by the adoption of attention mechanisms and temporal convolutions. The proposed method led to 81.65% and 80.72% classification accuracy for window sizes of 300 ms and 200 ms, respectively. The number of parameters to train the proposed TC-HGR architecture is 11.9 times less than that of its state-of-the-art counterpart.

Our understanding of the firing behaviour of motoneuron (MN) pools during human voluntary muscle contractions is currently limited to electrophysiological findings from animal experiments extrapolated to humans, mathematical models of MN pools not validated for human data, and experimental results obtained from decomposition of electromyographical (EMG) signals. These approaches are limited in accuracy or provide information on only small partitions of the MN population. Here, we propose a method based on the combination of high-density EMG (HDEMG) data and realistic modelling for predicting the behaviour of entire pools of motoneurons in humans. The method builds on a physiologically realistic model of a MN pool which predicts, from the experimental spike trains of a smaller number of individual MNs identified from decomposed HDEMG signals, the unknown recruitment and firing activity of the remaining unidentified MNs in the complete MN pool. The MN pool model is described as a cohort of single-compartment leaky fire-and-integrate (LIF) models of MNs scaled by a physiologically realistic distribution of MN electrophysiological properties and driven by a spinal synaptic input, both derived from decomposed HDEMG data. The MN spike trains and effective neural drive to muscle, predicted with this method, have been successfully validated experimentally. A representative application of the method in MN-driven neuromuscular modelling is also presented. The proposed approach provides a validated tool for neuroscientists, experimentalists, and modelers to infer the firing activity of MNs that cannot be observed experimentally, investigate the neuromechanics of human MN pools, support future experimental investigations, and advance neuromuscular modelling for investigating the neural strategies controlling human voluntary contractions.

Human–machine interaction requires accurate recognition of human intentions (e.g., via hand gestures). Here, we assessed the cross-day robustness of widely used hand gesture classification techniques applied to high-density surface electromyogram (HD-sEMG) signals (256 channels). Our evaluation covered techniques in each stage of the classification framework: first, 50 temporal-spectral-spatial domain features, second, 15 feature optimization techniques, and third, seven classifiers. Moreover, although HD-sEMG provides sufficient neuromuscular information, some of the channels may present low signal-to-noise ratio and should therefore be treated as outliers. Accordingly, we performed our evaluation with, first, all outlier channels retained, and second, removal of the features corresponding to poor-quality channels and substitution with interpolated values from neighbor channels. The impact of sliding window and data augmentation was also investigated. We examined the results on a 35-gesture classification task using HD-sEMG acquired from 20 subjects on two sessions in separate days. The results showed that interpolation of features from outlier channels significantly improved the performance in most cases. Use of a sliding window and of data augmentation contributed to a higher classification accuracy. For the classification of 11 selected gestures of common daily use, the support vector machine classifier achieved the highest classification accuracy of 91.9% in a cross-day validation protocol using an optimal combination of 13 features (each extracted from sliding windows), feature optimization by linear discriminant analysis, and data augmentation. Our work can serve as a technique-screening tool on cross-day applications of human–machine interactions.

Enhancing information security via reliable user authentication in wireless body area network (WBAN)-based Internet-of-Things (IoT) applications has attracted increasing attention. The noncancelability of traditional biometrics (e.g., fingerprint) for user authentication increases the privacy disclosure risks once the biometric template is exposed, because users cannot volitionally create a new template. In this work, we propose a cancelable biometric modality based on high-density surface electromyogram (HD-sEMG) encoded by hand gesture password, for user authentication. HD-sEMG signals (256 channels) were acquired from the forearm muscles when users performed a prescribed gesture password, forming their biometric token. Thirty four alternative hand gestures in common daily use were studied. Moreover, to reduce the data acquisition and transmission burden in IoT devices, an automatically generated password-specific channel mask was employed to reduce the number of active channels. HD-sEMG biometrics were also robust with reduced sampling rate, further reducing power consumption. HD-sEMG biometrics achieved a low equal error rate (EER) of 0.0013 when impostors entered a wrong gesture password, as validated on 20 subjects. Even if impostors entered the correct gesture password, the HD-sEMG biometrics still achieved an EER of 0.0273. If the HD-sEMG biometric template was exposed, users could cancel it by simply changing it to a new gesture password, with an EER of 0.0013. To the best of our knowledge, this is the first study to employ HD-sEMG signals under common daily hand gestures as biometric tokens, with training and testing data acquired on different days.

Dexterous control of upper limb prostheses with multiarticulated wrists/hands is still a challenge due to the limitations of myoelectric man-machine interfaces. Multiple factors limit the overall performance and usability of these interfaces, such as the need to control degrees of freedom sequentially and not concurrently, and the inaccuracies in decoding the user intent from weak or fatigued muscles. In this article, we developed a novel man-machine interface that endows a myoelectric prosthesis (MYO) with artificial perception, estimation of user intention, and intelligent control (MYO-PACE) to continuously support the user with automation while preparing the prosthesis for grasping. We compared the MYO-PACE against state-of-the-art myoelectric control (pattern recognition) in laboratory and clinical tests. For this purpose, eight able-bodied and two amputee individuals performed a standard clinical test consisting of a series of manipulation tasks (portion of the SHAP test), as well as a more complex sequence of transfer tasks in a cluttered scene. In all tests, the subjects not only completed the trials faster using the MYO-PACE but also achieved more efficient myoelectric control. These results demonstrate that the implementation of advanced perception, context interpretation, and autonomous decision-making into active prostheses improves control dexterity. Moreover, it also effectively supports the user by speeding up the preshaping phase of the movement and decreasing muscle use.

The rectified surface EMG signal is commonly used as an estimator of the neural drive to muscles and therefore to infer sources of synaptic input to motor neurons. Loss of EMG amplitude due to the overlap of motor unit action potentials (amplitude cancellation), however, may distort the spectrum of the rectified EMG and thereby its correlation with the neural drive. In this study, we investigated the impact of amplitude cancelation on this correlation using analytical derivations and a computational model of motor neuron activity, force, and the EMG signal. First, we demonstrated analytically that an ideal rectified EMG signal without amplitude cancellation (EMGnc) is superior to the actual rectified EMG signal as estimator of the neural drive to muscle. This observation was confirmed by the simulations, as the average coefficient of determination (r2) between the neural drive in the 1-30 Hz band and EMGnc (0.59±0.08) was matched by the correlation between the rectified EMG and the neural drive only when the level of amplitude cancellation was low (<40%) at low contraction levels (<5% of maximum voluntary contraction force; MVC). This correlation, however, decreased linearly with amplitude cancellation (r = -0.83) to values of r2 <0.2 at amplitude cancellation levels >60% (contraction levels >15% MVC). Moreover, the simulations showed that a stronger (i.e. more variable) neural drive implied a stronger correlation between the rectified EMG and the neural drive and that amplitude cancellation distorted this correlation mainly for low-frequency components (<5 Hz) of the neural drive. In conclusion, the results indicate that amplitude cancellation distorts the spectrum of the rectified EMG signal. This implies that valid use of the rectified EMG as an estimator of the neural drive requires low contraction levels and/or strong common synaptic input to the motor neurons.

Closing the prosthesis control loop by providing tactile sensory feedback to the user is a key point in research on active prosthetics as well as an often cited requirement of the prosthesis users. The live demo system includes: 1) electronic skin (e -skin) including a matrix of 64 sensing elements (taxels), 2) interface electronics for signal conditioning and data acquisition, 3) fully programmable multichannel electrotactile stimulator connected to flexible electrode matrices, and 4) host laptop PC which runs the online control loop implemented in Matlab.

Electronic skin can be integrated into a prosthetic device to endow the prosthesis with artificial cutaneous sensing, thereby partially restoring the sensory information lost due to an amputation. Non-invasive cutaneous electrostimulation transmits the tactile information sensed by the electronic skin on the prosthetic hand to the human brain, through the amputee's afferent nervous system. In this paper, our current benchtop prototype of a distributed sensing-stimulation system is presented, together with the envisaged high-fidelity solution which will be integrated into a real prosthetic hand.

The neural command from motor neurons to muscles - sometimes referred to as the neural drive to muscle - can be identified by decomposition of electromyographic (EMG) signals. This approach can be used for inferring the voluntary commands in neural interfaces in patients with limb amputations. This paper proposes for the first time an innovative method for fully automatic and real-time intramuscular EMG (iEMG) decomposition. The method is based on online single-pass density-based clustering and adaptive classification of bivariate features, using the concept of potential measure. No attempt was made to resolve superimposed motor unit action potentials. The proposed algorithm was validated on sets of simulated and experimental iEMG signals. Signals were recorded from the biceps femoris long-head, vastus medialis and lateralis and tibialis anterior muscles during low-to-moderate isometric constant-force and linearly-varying force contractions. The average number of missed, duplicated and erroneous clusters for the examined signals was [Formula: see text], [Formula: see text], and [Formula: see text], respectively. The average decomposition accuracy (defined similar to signal detection theory but without using True Negatives in the denominator) and coefficient of determination (variance accounted for) for the cumulative discharge rate estimation were [Formula: see text], and [Formula: see text], respectively. The time cost for processing each 200[Formula: see text]ms iEMG interval was [Formula: see text] (21-97)[Formula: see text]ms. However, computational time generally increases over time as a function of frames/signal epochs. Meanwhile, the incremental accuracy defined as the accuracy of real-time analysis of each signal epoch, was [Formula: see text]% for epochs recorded after initial one second. The proposed algorithm is thus a promising new tool for neural decoding in the next-generation of prosthetic control.

Tactile feedback is pivotal for grasping and manipulation in humans. Providing functionally effective sensory feedback to prostheses users is an open challenge. Past paradigms were mostly based on vibro- or electrotactile stimulations. However, the tactile sensitivity on the targeted body parts (usually the forearm) is greatly less than that of the hand/fingertips, restricting the amount of information that can be provided through this channel. Visual feedback is the most investigated technique in motor learning studies, where it showed positive effects in learning both simple and complex tasks; however, it was not exploited in prosthetics due to technological limitations. Here, we investigated if visual information provided in the form of augmented reality (AR) feedback can be integrated by able-bodied participants in their sensorimotor control of a pick-and-lift task while controlling a robotic hand. For this purpose, we provided visual continuous feedback related to grip force and hand closure to the participants. Each variable was mapped to the length of one of the two ellipse axes visualized on the screen of wearable single-eye display AR glasses. We observed changes in behavior when subtle (i.e., not announced to the participants) manipulation of the AR feedback was introduced, which indicated that the participants integrated the artificial feedback within the sensorimotor control of the task. These results demonstrate that it is possible to deliver effective information through AR feedback in a compact and wearable fashion. This feedback modality may be exploited for delivering sensory feedback to amputees in a clinical scenario.

Motor units are the smallest functional units of our movements. The study of their activation provides a window into the mechanisms of neural control of movement in humans. The classic methods for motor unit investigations date to several decades ago. They are based on invasive recordings with selective needle or wire electrodes. Conversely, the noninvasive (surface) EMG has been commonly processed as an interference signal, with the extraction of its global characteristics, e.g., amplitude. These characteristics, however, are only crudely associated to the underlying motor unit activities. In the last decade, methods have been proposed for reliably extracting individual motor unit activities from the interference surface EMG signal. We describe these methods in this review, with a focus on blind source separation (BSS) and techniques used on decomposed EMG signals. For example, from the motor unit discharge timings, information can be extracted regarding the synaptic input received by the corresponding motor neurons. In reviewing these methods, we also provide examples of applications in representative conditions, such as pathological tremor. In conclusion, we provide an overview of processing methods of the surface EMG signal that allow a reliable characterization of individual motor units in vivo in humans.

Myoelectric control has seen decades of research as a potential interface between human and machines. High-density surface electromyography (HDsEMG) non-invasively provides a rich set of signals representing underlying muscle contractions and, at a higher level, human motion intent. Many pattern recognition techniques have been proposed to predict motions based on these signals. However, control schemes incorporating pattern recognition struggle with long-term reliability due to signal stochasticity and transient changes. This study proposes an alternative approach for HDsEMG-based interfaces using concepts of motor skill learning and muscle synergies to address long-term reliability. Muscle synergy-inspired decomposition reduces HDsEMG into control inputs robust to small electrode displacements. The novel control scheme provides simultaneous and proportional control, and is learned by the subject simply by interacting with the device. In a multiple-day experiment, subjects learned to control a virtual 7-DoF myoelectric interface, displaying performance learning curves consistent with motor skill learning. On a separate day, subjects intuitively transferred this learning to demonstrate precision tasks with a 7-DoF robot arm, without requiring any recalibration. These results suggest that the proposed method may be a practical alternative to pattern recognition-based control for long-term use of myoelectric interfaces.

The paper presents an online estimation of parameters of a multi-input renewal Markov process. The underlying model is derived from the physiological generation of intramuscular electromyographic (iEMG) signals, which are recorded by wire electrodes. The iEMG is the sum of several sparse spikes trains and noise. An hidden Markov model, whose parameters express the muscular activity, is developed. The time duration between spikes is modeled with a discrete Weibull distribution, helping us to reduce the complexity of the estimation done with the help of a Bayes filter.

Diagnosis of bruxism is difficult since not all contractions of masticatory muscles during sleeping are bruxism episodes. In this paper, we propose the use of both EMG and ECG signals for the detection of sleep bruxism. Data have been acquired from 21 healthy volunteers and 21 sleep bruxers. The masseter surface EMGs were detected with bipolar concentric electrodes and the ECG with monopolar electrodes located on the clavicular regions. Recordings were made at the subjects' homes during sleeping. Bruxism episodes were automatically detected as characterized by masseter EMG amplitude greater than 10% of the maximum and heart rate increasing by more than 25% with respect to baseline within 1 s before the increase in EMG amplitude above the 10% threshold. Furthermore, the subjects were classified as bruxers and nonbruxers by a neural network. The number of bruxism episodes per night was 24.6 ± 8.4 for bruxers and 4.3 ± 4.5 for controls ( P < 0.0001). The classification error between bruxers and nonbruxers was 1% which was substantially lower than when using EMG only for the classification. These results show that the proposed system, based on the joint analysis of EMG and ECG, can provide support for the clinical diagnosis of bruxism.

Intra-cortical signals are usually affected by high levels of noise (0 dB SNR is not uncommon) either due to the recording equipment or to magnetical and electrical couplings between surrounding sources and the recording system. Besides from hindering effective exploitation of the information content in the signals, noise also influences the bandwidth needed to transmit them, which is a problem especially when a large number of channels are to be recorded. In this paper we propose a novel technique for joint denoising and compression of intra-cortical signals based on the Minimum Description Length principle (MDL). This method was tested on simulated signals and the results showed that the proposed technique achieves improvements in SNR (up to .6 dB over MNML for very noisy signals) and compression ratios greater than alternative denoising/compression methods.

Tremor is the most prevalent movement disorder, and its incidence is increasing with aging. In spite of the numerous therapeutic solutions available, 65% of those suffering from upper limb tremor report serious difficulties during their daily living. This gives rise to research on different treatment alternatives, amongst which wearable robots that apply selective mechanical loads constitute an appealing approach. In this context, the current work presents a multimodal human-robot interface to drive a neuroprosthesis for tremor management. Our approach relies on the precise characterization of the tremor to modulate a functional electrical stimulation system that compensates for it. The neuroprosthesis is triggered by the detection of the intention to move derived from the analysis of electroencephalographic activity, which provides a natural interface with the user. When a prediction is delivered, surface electromyography serves to detect the actual onset of the tremor in the presence of volitional activity. This information in turn triggers the stimulation, which relies on tremor parameters-amplitude and frequency-derived from a pair of inertial sensors that record the kinematics of the affected joint. Surface electromyography also yields a first characterization of the tremor, together with precise information on the preferred stimulation site. Apart from allowing for an optimized performance of the system, our multimodal approach permits the implementation of redundant methods to both enhance the reliability of the system and adapt to the specific needs of different users. Results with a representative group of patients illustrate the performance of the interface presented here and demonstrate its feasibility.

Tremor constitutes the most common motor disorder, and poses a functional problem to a large number of patients. Despite of the considerable experience in tremor management, current treatment based on drugs or surgery does not attain an effective attenuation in 25 % of patients, motivating the need for research in new therapeutic alternatives. In this context, this paper presents the concept design, development, and preliminary validation of a soft wearable robot for tremor assessment and suppression. The TREMOR neurorobot comprises a Brain Neural Computer Interface that monitors the whole neuromusculoskeletal system, aiming at characterizing both voluntary movement and tremor, and a Functional Electrical Stimulation system that compensates for tremulous movements without impeding the user perform functional tasks. First results demonstrate the performance of the TREMOR neurorobot as a novel means of assessing and attenuating pathological tremors.

During the last decade there has been a renewed interest in the development of advanced, active hand prosthetic devices for amputees. In contrast to passive prostheses, active devices can be controlled by the user's intention. Active prosthetic devices have been substantially improved by integrating robot technology to achieve more functionalities and lifelike movements. Despite important progress in the technological development of prosthetics, their clinical application is still limited by the quantity and quality of biological signals that can be used for understanding the user's intention, and by the relatively poor performance of the algorithms that translate the user's intention into a desired movement. In this review we describe a solution to some of these limitations, i.e., the flexible, multichannel, implantable intraneural and intramuscular electrodes to interface the body's peripheral nerves or muscles. We aim to review the historic development, the underlying technology, and the design concepts of these electrodes. Moreover, the signal processing methods applied to these recordings and their use for the control of prosthetic devices will be discussed. Although the focus is on hand prostheses, the interface approach described is general.

We propose a linear predictive coding technique for multichannel electromyographic (EMG) recordings. The signals are acquired using two-dimensional grid of electrodes which generate strongly correlated signals. Previous work only considered spectral redundancy across the signal matrix. In this paper we exploit the correlation present in the residual signals, i.e., the signals after the short term prediction. The proposed technique achieves a compression ratio of about 1divide9, i.e., slightly better than spectral-only decorrelation methods, but with a strong increase of approximately 3.2 dB SNR in the quality of the reconstructed waveform.

In this paper we extend a lossy compression technique for surface EMG signals, which is based on the algebraic code excited linear prediction (ACELP) paradigm, to compress multi-channel surface EMG recordings by exploiting the correlation between the line spectral frequencies (LSF). Experimental results show that the LSFs of the inner signals in a multi-channel recording can be efficiently represented with 13 bit/frame, versus the 38 bit/frame needed by independent ACELP coding of each signal, thus saving 66% of the bandwidth needed to transmit these coefficients while maintaining comparable performance in terms of the SNR, average rectified value and root mean square of the waveform, and mean and median frequencies of the power spectrum.

In this paper we investigate a lossy coding technique for surface EMG signals which is based on the algebraic code excited linear prediction (ACELP) paradigm, widely used for speech signal coding. The algorithm was adapted to the EMG characteristics and tested on both simulated and experimental signals. A fixed compression ratio of 87.3% was chosen. On simulated signals, the mean square error in signal reconstruction and the percentage error in average rectified value after compression were 10.43 % and 5.52 %, respectively. On experimental signals, they were 6.74% and 3.11%. The mean power spectral frequency and third order power spectral moment were estimated with relative error smaller than 1.36% and 1.70%, respectively, for simulated signals, and 3.74% and 2.28% for experimental signals. It was concluded that the proposed coding scheme can be effectively used for high rate, low distortion and low-delay compression of surface EMG signals

In this work, we propose a novel scheme of signal compression based on signal-dependent wavelets. To adapt the mother wavelet to the signal for the purpose of compression, it is necessary to define a family of wavelets that depend on a set of parameters and a quality criterion for wavelet selection (i.e., wavelet parameter optimization). We propose the use of orthogonal wavelets parameterized by their scaling filter, with optimization criterion based on the minimization of signal distortion rate given the desired compression rate. For coding the wavelet coefficients we adopted the embedded zerotree wavelet coding algorithm. Results on electromyographic signals show that optimization significantly improves performance.

Despite the great interest towards long term recordings of electromyographic (EMG) signals, which find applications, for example, in telemedicine, only a few studies have dealt with the compression of these signals. We propose a lossy coding technique for surface EMG signals. The technique is based on the linear predictive coding paradigm widely used for speech compression. The algorithm was tested on both simulated and experimental signals. Mean frequency, median frequency, variance, skewness and kurtosis of the EMG signals were preserved with an error less than 3% with respect to the original values for synthetic signals and experimental signals, reducing the bitrate from 24 kbit/s (12 kbit/s after downsampling) to 352 bit/s, with a compression factor of 97.1%. It was concluded that the linear predictive coding paradigm can be effectively used for high rate compression of surface EMG signals when preservation of only the power spectrum of the signal is of interest. This has applications in ergonomics and occupational medicine.