Electromyography Signals Research Articles

A multi-resolution physics-informed recurrent neural network: formulation and application to musculoskeletal systems.

This work presents a multi-resolution physics-informed recurrent neural network (MR PI-RNN), for simultaneous prediction of musculoskeletal (MSK) motion and parameter identification of the MSK systems. The MSK application was selected as the model problem due to its challenging nature in mapping the high-frequency surface electromyography (sEMG) signals to the low-frequency body joint motion controlled by the MSK and muscle contraction dynamics. The proposed method utilizes the fast wavelet transform to decompose the mixed frequency input sEMG and output joint motion signals into nested multi-resolution signals. The prediction model is subsequently trained on coarser-scale input-output signals using a gated recurrent unit (GRU), and then the trained parameters are transferred to the next level of training with finer-scale signals. These training processes are repeated recursively under a transfer-learning fashion until the full-scale training (i.e., with unfiltered signals) is achieved, while satisfying the underlying dynamic equilibrium. Numerical examples on recorded subject data demonstrate the effectiveness of the proposed framework in generating a physics-informed forward-dynamics surrogate, which yields higher accuracy in motion predictions of elbow flexion-extension of an MSK system compared to the case with single-scale training. The framework is also capable of identifying muscle parameters that are physiologically consistent with the subject's kinematics data.

Automatic selection of IMFs to denoise the sEMG signals using EMD

Surface Electromyography (sEMG) signals are muscle activation signals, which has applications in muscle diagnosis, rehabilitation, prosthetics, and speech etc. However, they are known to be affected by noises such as Power Line Interference (PLI), motion artifacts etc. Currently, Empirical Mode Decomposition (EMD) and its modifications such as Ensemble EMD (EEMD), and Complementary EEMD (CEEMD) are used to decompose EMG into a series of Intrinsic Mode Functions (IMFs). The denoised EMG can be obtained from the selected IMFs. Statistical methods are used to select the signal dominant IMFs to reconstruct the denoised signal. In this work, a novel procedure is proposed to automatically separate noisy IMFs from the original sEMG signal. For this purpose, Permutation Entropy (PE) is employed in EEMD sifting process called Partly EEMD (PEEMD), to separate the noisy IMFs from the original sEMG signal according to the preset PE threshold. PEEMD decomposes the original signal into various modes according to a preset PE threshold and the denoised signal is reconstructed from resultant IMFs. The PEEMD denoising procedure is applied on the experimental sEMG data collected from eight subjects, that include six various upper limb movement classes. The proposed denoising procedure achieved an improved denoising performance in comparison with EMD, EEMD, and CEEMD. An alternate measure called Sample Entropy (SE) is also used in place of PE, for the automated sifting process as a comparison. Signal to Noise Ratio (SNR), Root Mean Square Error (RMSE), and Reconstruction Error (RE) parameters are used to evaluate the denoising performance. The results, averaged across eight subjects, demonstrate that the proposed denoising procedure outperforms the state-of-the-art EMD techniques in terms of these performance measures on the experimentally collected sEMG data samples.

Effects of muscle strength in different parts of adolescent standing long jump on distance based on surface electromyography.

Objective: To reveal the influence of muscle strength in different parts of the body on the distance of standing jump, to establish the key force phases of muscle strength in different parts, and to improve the recognition of movement norms. Methods: VICON infrared three-dimensional motion capture acquisition and analysis system and Noraxon Ultium surface electromyography acquisition and analysis system were used to complete the surface electromyography signal acquisition of 18 randomly selected subjects performing standing long jump. Results: 1) Triceps brachii, anterior deltoid, latissimus dorsi, semitendinosus, rectus femoris, upper trapezius, pectoralis major, and biceps femoris had significant effects on standing jump distance. 2) From the point of view of the key exertion phase of the standing jump mainly affecting the muscle group; the main exertion phase of the semitendinosus occurs from the rising stage to the descending stage; the rectus femoris, triceps, and latissimus dorsi occur during the ascending phase of the flight; the anterior deltoid muscle occurs in the transition stage from rising to falling in the air; the trapezius muscle occurs in the transition stage from pre-swing to kick-off. Conclusion: 1) From the regression analysis of the measured muscles on the distance of each stage of standing long jump, deltoid muscle strength is conducive to the improvement of standing long jump distance, which further indicates the importance of upper limb deltoid muscle strength. 2) Through time series analysis, it is found that the force performance of the rectus femoris muscle at this stage can be used as the main identification parameter of standing long jump, and can effectively distinguish different types of movements.

Design and evaluation of a parallel mechanism for wearable lumbar support exoskeleton.

Work-related musculoskeletal disorders (WMSDs) are a serious problem, and manual material handling (MMH) tasks remain common in most industries. Thus, a lightweight and active exoskeleton is to be needed. A facile, convenient, multifunctional, wearable lumbar support exoskeleton (WLSE) was proposed to relieve the muscular tension and fatigue especially in the way of WMSDs. Based on the screw theory and virtual power principle, the parallel structure was used as the scheme choice for selecting suitable actuators and joints. The exoskeleton, which was characterized by high adaptability and complied with human motion, included branch unit, mechanism branch units, control units and sensors. Furthermore, using surface electromyography (sEMG) signal evaluation, an experiment which contains several tests were designed to evaluate whether WLSE had effect on supporting and reliving muscular fatigue while lifting-up different weight of objects under wearing without traction (T1) and wearing with traction (T2). Data collected were analyzed statistically by the two-way ANOVA. It showed that the RMS of sEMG was obviously reduced while carrying the heavy objects with WLSE under T2, and the MF values always performed the decreasing trend in T2/T1. This paper proposed a facile, convenient, multifunctional WLSE. From the results, it was concluded that the WLSE was significantly effective in reliving the muscle tension and muscle fatigue while lifting to prevent and treat WMSDs.

The relationship between EMG high frequency and low frequency band amplitude changes correlates with tissue inorganic phosphate levels.

Assessing inorganic phosphate levels seems crucial in deciphering the biochemical state of organisms or tissues. The concentration of inorganic phosphate in blood is an order of magnitude smaller than in tissues and, on top of that, it is dynamically used to fill temporary gaps in tissues. This is the reason blood inorganic phosphate level is considered a poor proxy for tissue levels. Therefore, tissue biopsy seems to be the dominant method when assessing inorganic phosphate levels for instance in muscles. In this study, we attempted to derive a non-invasive biomarker for phosphate tissue levels. We analyzed surface electromyography signals taken during 31P spectroscopy of leg muscles in five adult pigs. We induced hypophosphatemia via 20 minutes-long hyperventilation. It turned out that the proportion of the amplitude of the low frequency band and the high frequency band is significantly (p=0.002) correlated with the relative phosphate levels. The electromyographic signal did not correlate significantly with pCO2 levels in the blood, suggesting that the changes in the signal are a result of inorganic phosphate levels, not hyperventilation. The results might lead to the development of a real-time phosphate fluctuations measurement procedure.

Changes in neuromuscular activation, heart rate and rate of perceived exertion over the course of a wheelchair propulsion fatigue protocol.

Shoulder pain is common in persons with spinal cord injury and has been associated with wheelchair use. Fatigue related compensation strategies have been identified as possibly impacting the development of shoulder injury and pain. The purpose of this study was to investigate the progression of performance fatigability (i.e., decline in objective measure of performance including neuromuscular activation and increase in heart rate) and perceived fatigability (i.e., increased perceived exertion) during a 15-min fatigue protocol including maximum voluntary overground wheelchair propulsion. Fifty participants with paraplegic spinal cord injury completed three 4-min rounds of wheelchair propulsion, separated by 90s of rest, on a figure-8 course consisting of two turns and full stops per lap in their manual wheelchairs (ClinicalTrials.gov: NCT03153033). Electromyography (EMG) signal of five muscles acting on the shoulder joint, heart rate (HR), and rate of perceived exertion (RPE) were measured at the beginning and end of every 4min of propulsion. Root Mean Square (RMS) and Mean Power Frequency were calculated from EMG data. There was a significant increase in %RMS of the pectoralis major pars sternalis and trapezius pars descendens, HR, and RPE with greatest changes during the first 4min of the protocol. The observed changes in neuromuscular activation in only two of the shoulder muscles may impact muscular imbalances and the development of shoulder injuries and should be further studied. The current study gives clearer insight into the mechanisms of performance fatigability and perceived fatigability throughout a wheelchair propulsion fatigue protocol.

Skin-integrated, biocompatible, and stretchable silicon microneedle electrode for long-term EMG monitoring in motion scenario

Electromyography (EMG) signal is the electrical potential generated by contracting muscle cells. Long-term and accurate EMG monitoring is desirable for neuromuscular function assessment in clinical and the human–computer interfaces. Herein, we report a skin-integrated, biocompatible, and stretchable silicon microneedle electrode (SSME) inspired by the plant thorns. The silicon microneedles are half encapsulated by the polyimide (PI) to enhance the adaptability to deformation and resistance to fatigue. Thorn-like SSME is realized by the semi-additive method with a stretchability of not less than 36%. The biocompatibility of SSME has been verified using cytotoxicity tests. EMG monitoring in motion and long-term has been conducted to demonstrate the feasibility and performance of the SSME, which is compared with a commercial wet electrode. Hopefully, the strategies reported here can lead to accurate and long-term EMG monitoring, facilitating an effective and reliable human–computer interface.

Muscle synergies in joystick manipulation.

Extracting muscle synergies from surface electromyographic signals (sEMGs) during exercises has been widely applied to evaluate motor control strategies. This study explores the relationship between upper-limb muscle synergies and the performance of joystick manipulation tasks. Seventy-seven subjects, divided into three classes according to their maneuvering experience, were recruited to perform the left and right reciprocation of the joystick. Based on the motion encoder data, their manipulation performance was evaluated by the mean error, standard deviation, and extreme range of position of the joystick. Meanwhile, sEMG and acceleration signals from the upper limbs corresponding to the entire trial were collected. Muscle synergies were extracted from each subject's sEMG data by non-negative matrix factorization (NMF), based on which the synergy coordination index (SCI), which indicates the size of the synergy space and the variability of the center of activity (CoA), evaluated the temporal activation variability. The synergy pattern space and CoA of all participants were calculated within each class to analyze the correlation between the variability of muscle synergies and the manipulation performance metrics. The correlation level of each class was further compared. The experimental results evidenced a positive correlation between manipulation performance and maneuvering experience. Similar muscle synergy patterns were reflected between the three classes and the structure of the muscle synergies showed stability. In the class of rich maneuvering experience, the correlation between manipulation performance metrics and muscle synergy is more significant than in the classes of trainees and newbies, suggesting that long-term training and practicing can improve manipulation performance, stability of synergy compositions, and temporal activation variability but not alter the structure of muscle synergies determined by a specific task. Our approaches and findings could be applied to 1) reduce manipulation errors, 2) assist maneuvering training and evaluation to enhance transportation safety, and 3) design technical support for sports.

Self-healing electrical bioadhesive interface for electrophysiology recording

Electrical bioadhesive interfaces (EBIs) are standing out in various applications, including medical diagnostics, prosthetic devices, rehabilitation, and human–machine interactions. Nonetheless, crafting a reliable and advanced EBI with comprehensive properties spanning electrochemical, electrical, mechanical, and self-healing capabilities remains a formidable challenge. Herein, we develop a self-healing EBI by thoughtfully integrating conducting polymer nanofibers and a typical bioadhesive within a robust hydrogel matrix. The accomplished EBI demonstrates extraordinary adhesion (lap shear strength of 197 kPa), exceptional electrical conductivity (2.18 S m−1), and outstanding self-healing performance. Taking advantage of these attributes, we integrated the EBI into flexible skin electrodes for surface electromyography (sEMG) signal recording from forearm muscles. The engineered skin electrodes exhibit robust adhesion to the skin even when sweating, rapid self-healing from damage, and seamless real-time signal recording with a higher signal-to-noise ratio (39 dB). Our EBI, along with its skin electrodes, offers a promising platform for tissue-device integration, health monitoring, and an array of bioelectronic applications.

Characteristics of multi-channel intermuscular directional coupling based on time-varying partial directional coherence analysis

The human body transmits directional information between muscles during upper limb movements, and this will be particularly evident when the dominant muscle changes during movement transitions. By capturing the electromyography (EMG) signals of wrist flexion and extension continuous transition movements, we investigated the characteristics of multichannel intermuscular directional coupling and directional information transmission, and consequently explored the control mechanism of Central nervous system (CNS) and the coordination mechanism of motor muscles. Multi-channel EMG was collected from 12 healthy subjects under continuous translational movements of wrist flexion and extension, and the time-varying biased directional coherence analysis (TVPDC) model was constructed using partial directional coherence analysis (PDC) frequency domain directionality to study the directional information transfer characteristics in the time–frequency domain, screen closely related muscle pairs and perform directional coupling significance analysis. Palmaris longus (PL) played a dominant role under wrist flexion movements(WF), Extensor Carpi Radialis (ECR) played a dominant role under wrist extension movements(WE), and the remaining muscles responded to them with information and Biceps Brachii (BB) played a responsive role throughout the movement; flexor pairs had the highest positive coupling values in the beta band during Conversion action1 (MC1) and WF phases, and extensor pairs had the highest positive coupling values in the gamma band during Conversion action2(MC2) phase and the highest coupling values in the beta band during WE phase. TVPDC can effectively analyze the multichannel intermuscular directional coupling and information transmission relationship of surface electromyography under wrist flexion and extension transition movements, providing a reference for exploring the control mechanism of CNS and abnormal control mechanism in patients with motor dysfunction in a new perspective.

A Smart, Textile-Driven, Soft Exosuit for Spinal Assistance.

Work-related musculoskeletal disorders (WMSDs) are often caused by repetitive lifting, making them a significant concern in occupational health. Although wearable assist devices have become the norm for mitigating the risk of back pain, most spinal assist devices still possess a partially rigid structure that impacts the user's comfort and flexibility. This paper addresses this issue by presenting a smart textile-actuated spine assistance robotic exosuit (SARE), which can conform to the back seamlessly without impeding the user's movement and is incredibly lightweight. To detect strain on the spine and to control the smart textile automatically, a soft knitting sensor that utilizes fluid pressure as a sensing element is used. Based on the soft knitting hydraulic sensor, the robotic exosuit can also feature the ability of monitoring and rectifying human posture. The SARE is validated experimentally with human subjects (N = 4). Through wearing the SARE in stoop lifting, the peak electromyography (EMG) signals of the lumbar erector spinae are reduced by 22.8% ± 12 for lifting 5 kg weights and 27.1% ± 14 in empty-handed conditions. Moreover, the integrated EMG decreased by 34.7% ± 11.8 for lifting 5 kg weights and 36% ± 13.3 in empty-handed conditions. In summary, the artificial muscle wearable device represents an anatomical solution to reduce the risk of muscle strain, metabolic energy cost and back pain associated with repetitive lifting tasks.

Sign Language Recognition Using the Electromyographic Signal: A Systematic Literature Review.

The analysis and recognition of sign languages are currently active fields of research focused on sign recognition. Various approaches differ in terms of analysis methods and the devices used for sign acquisition. Traditional methods rely on video analysis or spatial positioning data calculated using motion capture tools. In contrast to these conventional recognition and classification approaches, electromyogram (EMG) signals, which measure muscle electrical activity, offer potential technology for detecting gestures. These EMG-based approaches have recently gained attention due to their advantages. This prompted us to conduct a comprehensive study on the methods, approaches, and projects utilizing EMG sensors for sign language handshape recognition. In this paper, we provided an overview of the sign language recognition field through a literature review, with the objective of offering an in-depth review of the most significant techniques. These techniques were categorized in this article based on their respective methodologies. The survey discussed the progress and challenges in sign language recognition systems based on surface electromyography (sEMG) signals. These systems have shown promise but face issues like sEMG data variability and sensor placement. Multiple sensors enhance reliability and accuracy. Machine learning, including deep learning, is used to address these challenges. Common classifiers in sEMG-based sign language recognition include SVM, ANN, CNN, KNN, HMM, and LSTM. While SVM and ANN are widely used, random forest and KNN have shown better performance in some cases. A multilayer perceptron neural network achieved perfect accuracy in one study. CNN, often paired with LSTM, ranks as the third most popular classifier and can achieve exceptional accuracy, reaching up to 99.6% when utilizing both EMG and IMU data. LSTM is highly regarded for handling sequential dependencies in EMG signals, making it a critical component of sign language recognition systems. In summary, the survey highlights the prevalence of SVM and ANN classifiers but also suggests the effectiveness of alternative classifiers like random forests and KNNs. LSTM emerges as the most suitable algorithm for capturing sequential dependencies and improving gesture recognition in EMG-based sign language recognition systems.

Improving Motion Intention Recognition for Trans-Radial Amputees Based on sEMG and Transfer Learning

Hand motion intentions can be detected by analyzing the surface electromyographic (sEMG) signals obtained from the remaining forearm muscles of trans-radial amputees. This technology sheds new light on myoelectric prosthesis control; however, fewer signals from amputees can be collected in clinical practice. The collected signals can further suffer from quality deterioration due to the muscular atrophy of amputees, which significantly decreases the accuracy of hand motion intention recognition. To overcome these problems, this work proposed a transfer learning strategy combined with a long-exposure-CNN (LECNN) model to improve the amputees’ hand motion intention recognition accuracy. Transfer learning can leverage the knowledge acquired from intact-limb subjects to amputees, and LECNN can effectively capture the information in the sEMG signals. Two datasets with 20 intact-limb and 11 amputated-limb subjects from the Ninapro database were used to develop and evaluate the proposed method. The experimental results demonstrated that the proposed transfer learning strategy significantly improved the recognition performance (78.1%±19.9%, p-value < 0.005) compared with the non-transfer case (73.4%±20.8%). When the source and target data matched well, the after-transfer accuracy could be improved by up to 8.5%. Compared with state-of-the-art methods in two previous studies, the average accuracy was improved by 11.6% (from 67.5% to 78.1%, p-value < 0.005) and 12.1% (from 67.0% to 78.1%, p-value < 0.005). This result is also among the best from the contrast methods.

The impact of a 40-min nap on neuromuscular fatigue profile and recovery following the 5-m shuttle run test.

This study aims to investigate the impact of a 40-min nap opportunity on perceived recovery, exertion, and maximal voluntary isometric contraction (MVIC) following the 5-m shuttle run test (5SRT), after 1 night of normal sleep. In a randomised, counterbalanced, cross-over design, 17 trained men (mean [SD] age 20 [3] years, height 173 [6] cm, body mass 68 [6] kg) performed a 5SRT under two conditions: a 40-min nap opportunity and no-nap condition. After both conditions, electromyography signals during a 5-s isometric knee extension were recorded before and immediately after the 5SRT. Two electrical nerve stimulations at the femoral nerve were measured during and after the MVIC. Force, voluntary activation level, M-wave amplitudes, potentiated twitch, and electromyography signals (root mean square) were measured during each MVIC. Perceived exertion was recorded after each repetition of the test and perceived recovery was determined after the end of the MVIC. Compared to the no-nap condition, the 40-min nap resulted in significant enhancements in both the highest distance (p < 0.01, Δ = +7.6%) and total distance (p < 0.01, Δ = +7.5%). Before and after exercise, values for MVIC, root mean square, M-wave amplitudes, and voluntary activation level were improved after the 40-min nap opportunity compared to no-nap condition (all p ≤ 0.01). Values for perceived exertion and recovery were improved after the 40-min nap opportunity in comparison with no-nap condition (p ≤ 0.01). A 40-min nap opportunity improved repeated high-intensity short-term maximal performance, perceived recovery, associated neuromuscular responses, and reduced perceived fatigue. Therefore, our findings suggest that central and peripheral processes are involved in the improvements of 5SRT performance after napping.

Muscle synergies analysis shows altered neural strategies in women with patellofemoral pain during walking.

Several studies suggest that the central nervous system coordinates muscle activation by modulating neural commands directed to groups of muscles combined to form muscle synergies. Individuals with patellofemoral pain (PFP) move differently from asymptomatic individuals. Understanding the neural strategies involved in the execution of tasks such as walking can help comprehend how the movement is planned and better understand this clinical condition. The objective of this study was to compare muscle synergies between women with and without PFP during walking. Eleven women with PFP and thirteen asymptomatic women were assessed using three-dimensional kinematics and electromyography (EMG) while walking at self-selected speed. Kinematics of the trunk, pelvis and lower limbs were analyzed through the Movement Deviation Profile. Muscle synergies were extracted from the EMG signals of eight lower limb muscles collected throughout the whole gait cycle. Kinematic differences between the two groups (p<0.001, z-score = 3.06) were more evident during loading response, terminal stance, and pre-swing. PFP group presented a lower number of muscle synergies (p = 0.037), and greater variability accounted for (VAFtotal) when using 3 (p = 0.017), 4 (p = 0.004), and 5 (p = 0.012) synergies to reconstruct all EMG signals. The PFP group also presented higher VAFmuscle for rectus femoris (p = 0.048) and gastrocnemius medialis (p = 0.019) when considering 4 synergies. Our results suggest that women with PFP show lower motor complexity and deficit in muscle coordination to execute gait, indicating that gait in PFP is the result of different neural commands compared to asymptomatic women.

A robust electromyography signals-based interaction interface for human-robot collaboration in 3D operation scenarios

At present, the serious challenge of human-robot collaboration is how to establish an interaction interface with multi-directional interactivity and robustness. This paper applies surface electromyography (sEMG) signals as the interface medium, and proposes a robust 3D arm strength estimation model based on hybrid deep learning networks (R3DNet) to build a stable interaction interface and achieve reliable human-robot collaboration performance in 3D operation scenarios. R3DNet adopts the modular design idea and includes a feature extraction module, a regression estimation module, and a robust enhancement module. In the feature extraction module, an autoencoder structure-based network and a new loss function are proposed to extract the fusion features of multi-channel sEMG signals. In the regression estimation module, an arm strength model based on transfer learning algorithms is proposed. This model considers the similarity of arm strength estimation in each direction and realizes the arm strength estimation from one to three dimensions through the fine-tune strategy. The robust enhancement module adopts a residual structure and aims to address the problem of poor model robustness due to non-ideal conditions such as electrode displacement, muscle fatigue, muscle time-varying characteristics and differences in muscle characteristics among subjects. Further, an improved adaptive admittance control method is proposed to help the robot to adjust collaborative behavior according to the changing arm strength of the human tutor. The results of comparison experiments and multi-point flexible assembly experiments verify the validity and engineering realizability of R3DNet for human-robot collaboration.

A rehabilitation robot control framework with adaptation of training tasks and robotic assistance.

Robot-assisted rehabilitation has exhibited great potential to enhance the motor function of physically and neurologically impaired patients. State-of-the-art control strategies usually allow the rehabilitation robot to track the training task trajectory along with the impaired limb, and the robotic motion can be regulated through physical human-robot interaction for comfortable support and appropriate assistance level. However, it is hardly possible, especially for patients with severe motor disabilities, to continuously exert force to guide the robot to complete the prescribed training task. Conversely, reduced task difficulty cannot facilitate stimulating patients' potential movement capabilities. Moreover, challenging more difficult tasks with minimal robotic assistance is usually ignored when subjects show improved performance. In this paper, a control framework is proposed to simultaneously adjust both the training task and robotic assistance according to the subjects' performance, which can be estimated from the users' electromyography signals. Concretely, a trajectory deformation algorithm is developed to generate smooth and compliant task motion while responding to pHRI. An assist-as-needed (ANN) controller along with a feedback gain modification algorithm is designed to promote patients' active participation according to individual performance variance on completing the training task. The proposed control framework is validated using a lower extremity rehabilitation robot through experiments. The experimental results demonstrate that the control scheme can optimize the robotic assistance to complete the subject-adaptation training task with high efficiency.

Volitional EMG Control Enables Stair Climbing with a Robotic Powered Knee Prosthesis.

Existing controllers for robotic powered prostheses regulate the prosthesis speed, timing, and energy generation using predefined position or torque trajectories. This approach enables climbing stairs step-over-step. However, it does not provide amputees with direct volitional control of the robotic prosthesis, a functionality necessary to restore full mobility to the user. Here we show that proportional electromyographic (EMG) control of the prosthesis knee torque enables volitional control of a powered knee prosthesis during stair climbing. The proposed EMG controller continuously regulates knee torque based on activation of the residual hamstrings, measured using a single EMG electrode located within the socket. The EMG signal is mapped to a desired knee flexion/extension torque based on the prosthesis knee position, the residual limb position, and the interaction with the ground. As a result, the proposed EMG controller enabled an above-knee amputee to climb stairs at different speeds, while carrying additional loads, and even backwards. By enabling direct, volitional control of powered robotic knee prostheses, the proposed EMG controller has the potential to improve amputee mobility in the real world.

Multivariable closed-loop control of deep brain stimulation for Parkinson’s disease

Objective. Closed-loop deep brain stimulation (DBS) methods for Parkinson’s disease (PD) to-date modulate either stimulation amplitude or frequency to control a single biomarker. While good performance has been demonstrated for symptoms that are correlated with the chosen biomarker, suboptimal regulation can occur for uncorrelated symptoms or when the relationship between biomarker and symptom varies. Control of stimulation-induced side-effects is typically not considered. Approach. A multivariable control architecture is presented to selectively target suppression of either tremor or subthalamic nucleus beta band oscillations. DBS pulse amplitude and duration are modulated to maintain amplitude below a threshold and avoid stimulation of distal large diameter axons associated with stimulation-induced side effects. A supervisor selects between a bank of controllers which modulate DBS pulse amplitude to control rest tremor or beta activity depending on the level of muscle electromyographic (EMG) activity detected. A secondary controller limits pulse amplitude and modulates pulse duration to target smaller diameter axons lying close to the electrode. The control architecture was investigated in a computational model of the PD motor network which simulated the cortico-basal ganglia network, motoneuron pool, EMG and muscle force signals. Main results. Good control of both rest tremor and beta activity was observed with reduced power delivered when compared with conventional open loop stimulation, The supervisor avoided over- or under-stimulation which occurred when using a single controller tuned to one biomarker. When DBS amplitude was constrained, the secondary controller maintained the efficacy of stimulation by increasing pulse duration to compensate for reduced amplitude. Dual parameter control delivered effective control of the target biomarkers, with additional savings in the power delivered. Significance. Non-linear multivariable control can enable targeted suppression of motor symptoms for PD patients. Moreover, dual parameter control facilitates automatic regulation of the stimulation therapeutic dosage to prevent overstimulation, whilst providing additional power savings.

Movement recognition via channel-activation-wise sEMG attention.

Surface electromyography (sEMG) signals contain rich information recorded from muscle movements and therefore reflect the user's intention. sEMG has seen dominant applications in rehabilitation, clinical diagnosis as well as human engineering, etc. However, current feature extraction methods for sEMG signals have been seriously limited by their stochasticity, transiency, and non-stationarity. Our objective is to combat the difficulties induced by the aforementioned downsides of sEMG and thereby extract representative features for various downstream movement recognition. We propose a novel 3-axis view of sEMG features composed of temporal, spatial, and channel-wise summary. We leverage the state-of-the-art architecture Transformer to enforce efficient parallel search and to get rid of limitations imposed by previous work in gesture classification. The transformer model is designed on top of an attention-based module, which allows for the extraction of global contextual relevance among channels and the use of this relevance for sEMG recognition. We compared the proposed method against existing methods on two Ninapro datasets consisting of data from both healthy people and amputees. Experimental results show the proposed method attains the state-of-the-art (SOTA) accuracy on both datasets. We further show that the proposed method enjoys strong generalization ability: a new SOTA is achieved by pretraining the model on a different dataset followed by fine-tuning it on the target dataset.