A new strategy for multifunction myoelectric control

Classification of EMG signals with CNN features and voting ensemble classifier

Electromyography (EMG) signals are primarily used to control prosthetic hands. Classifying hand gestures efficiently with EMG signals presents numerous challenges. In addition to overcoming these challenges, a successful combination of feature extraction and classification approaches will improve classification accuracy. In the current work, convolutional neural network (CNN) features are used to reduce the redundancy problems associated with time and frequency domain features to improve classification accuracy. The features from the EMG signal are extracted using a CNN and are fed to the 'k' nearest neighbor (KNN) classifier with a different number of neighbors ( 1 N N , 3 N N , 5 N N , and   7 N N ) . It results in an ensemble of classifiers that are combined using a hard voting-based classifier. Based on the benchmark Ninapro DB4 database and CapgMyo database, the proposed framework obtained 91.3 % classification accuracy on CapgMyo and 89.5 % on Ninapro DB4.

APOE- ε4 Modulates Facial Neuromuscular Activity in Nondemented Adults: Toward Sensitive Speech-Based Diagnostics for Alzheimer's Disease

The APOE-ε4 allele is a genetic risk factor for late-onset Alzheimer's disease (AD). Beyond cognitive decline, APOE-ε4 affects motor function, reducing muscle strength and coordination, potentially through mitochondrial dysfunction and oxidative stress. This study examined the influence of the APOE- ε4 allele on neuromuscular function in oral muscles involved in speech production, using surface electromyography (EMG); and assessed the predictive power of EMG measures in differentiating APOE- ε4 carriers from noncarriers. Forty-two cognitively intact adults (16 APOE- ε4 carriers, 26 noncarriers) completed speech tasks while EMG was recorded from seven craniofacial muscles. Seventy EMG features including amplitude, frequency, complexity, regularity, and functional connectivity were extracted. Statistical analyses assessed genotype effects, sex differences, and correlations with blood metabolic biomarkers. APOE- ε4 carriers exhibited increased motor unit recruitment and synchronization, suggesting accelerated muscle fatigue. EMG-based measures outperformed cognitive tests in distinguishing carriers (AUC = 0.90) and correlated with metabolic biomarkers. Sex differences emerged, with female carriers showing reduced and male carriers showing increased functional connectivity. These findings highlight speech-based neuromuscular changes as potential early biomarkers of Alzheimer's risk before cognition is affected.

First-in-human implementation of a bidirectional somatosensory neuroprosthetic system with wireless communication

BACKGROUND

Limitations in upper limb prosthesis function and lack of sensory feedback are major contributors to high prosthesis abandonment rates. Peripheral nerve stimulation and intramuscular recording can restore touch and relay motor intentions for individuals with upper limb loss. Percutaneous systems have enabled significant progress in implanted neural interfaces but require chronic lead maintenance and unwieldy external equipment. Fully implanted sensorimotor systems without percutaneous leads are crucial for advancing implanted neuroprosthetic technologies to long-term community use and commercialization.

METHODS

We present the first-in-human technical performance of the implanted Somatosensory Electrical Neurostimulation and Sensing (iSens®) system-an implanted, high-channel count myoelectric sensing and nerve stimulation system that uses wireless communication for advanced prosthetic systems. Two individuals with unilateral transradial amputations received iSens® with four 16-channel composite Flat Interface Nerve Electrodes (C-FINEs) and four Tetra Intramuscular (TIM) electrodes. This study achieved two key objectives to demonstrate system feasibility prior to long-term community use: (1) evaluating the chronic stability of extraneural cuff electrodes, intramuscular electrodes, and active implantable devices in a wirelessly connected system and (2) assessing the impacts of peripheral nerve stimulation on three degree-of-freedom controller performance in a wirelessly connected system to validate iSens® as a bidirectional interface.

RESULTS

Similar to prior percutaneous systems, we demonstrate chronically stable extraneural cuff electrodes and intramuscular electrodes in a wirelessly connected implanted system for more than two years in one participant and four months in the second participant, whose iSens® system was explanted due to an infection of unknown origin. Using an artificial neural network controller trained on implanted electromyographic data collected during known hand movements, one participant commanded a virtual hand and sensorized prosthesis in 3 degrees-of-freedom. The iSens® system simultaneously produced stimulation for sensation while recording high resolution muscle activity for real-time control. Although restored sensation did not significantly improve initial trials of prosthetic controller performance, the participant reported that sensation was helpful for functional tasks.

CONCLUSIONS

This case series describes a wirelessly connected, bidirectional neuroprosthetic system with somatosensory feedback and advanced myoelectric prosthetic control that is ready for implementation in long-term home use clinical trials.

TRIAL REGISTRATION

ClinicalTrials.gov ID: NCT04430218, 2020-06-30.

Surface EMG Sensing and Granular Gesture Recognition for Rehabilitative Pouring Tasks: A Case Study

Surface electromyography (sEMG) non-invasively captures the electrical activity generated by muscle contractions, offering valuable insights into motion intentions. While sEMG has been widely applied to general gesture recognition in rehabilitation, there has been limited exploration of specific, intricate daily tasks, such as the pouring action. Pouring is a common yet complex movement requiring precise muscle coordination and control, making it an ideal focus for rehabilitation studies. This research proposes a granular computing-based deep learning approach utilizing ConvMixer architecture enhanced with feature fusion and granular computing to improve gesture recognition accuracy. Our findings indicate that the addition of hand-crafted features significantly improves model performance; specifically, the ConvMixer model's accuracy improved from 0.9512 to 0.9929. These results highlight the potential of our approach in rehabilitation technologies and assistive systems for restoring motor functions in daily activities.

Gesture Recognition Through Mechanomyogram Signals: An Adaptive Framework for Arm Posture Variability

In hand gesture recognition, classifying gestures across multiple arm postures is challenging due to the dynamic nature of muscle fibers and the need to capture muscle activity through electrical connections with the skin. This paper presents a gesture recognition architecture addressing the arm posture challenges using an unsupervised domain adaptation technique and a wearable mechanomyogram (MMG) device that does not require electrical contact with the skin. To deal with the transient characteristics of muscle activities caused by changing arm posture, Continuous Wavelet Transform (CWT) combined with Domain-Adversarial Convolutional Neural Networks (DACNN) were used to extract MMG features and classify hand gestures. DACNN was compared with supervised trained classifiers and shown to achieve consistent improvement in classification accuracies over multiple arm postures. With less than 5 minutes of setup time to record 20 examples per gesture in each arm posture, the developed method achieved an average prediction accuracy of $87.43 \%$ for classifying 5 hand gestures in the same arm posture and $64.29 \%$ across 10 different arm postures. When further expanding the MMG segmentation window from $200 \,\mathrm{ms}$ to $600 \,\mathrm{ms}$ to extract greater discriminatory information at the expense of longer response time, the intra-posture and inter-posture accuracies increased to $92.32 \%$ and $71.75 \%$. The findings demonstrate the capability of the proposed method to improve generalization throughout dynamic changes caused by arm postures during non-laboratory usages and the potential of MMG to be an alternative sensor with comparable performance to the widely used electromyogram (EMG) gesture recognition systems.

Leveraging Extended Windows in End-to-End Deep Learning for Improved Continuous Myoelectric Locomotion Prediction

Current surface electromyography (sEMG) methods for locomotion mode prediction face limitations in anticipatory capability due to computation delays and constrained window lengths typically below 500 ms-a practice historically tied to stationarity requirements of handcrafted feature extraction. This study investigates whether end-to-end convolutional neural networks (CNNs) processing raw sEMG signals can overcome these constraints through extended window lengths (250 ms to 1500 ms). We systematically evaluate six window lengths paired with three prediction horizons (model forecasts 50 ms to 150 ms ahead) in a continuous locomotion task involving eight modes and 16 transitions. The optimal configuration (1000 ms window with 150 ms horizon) achieved subject-average accuracies of 96.93% (steady states) and 97.50% (transient states), maintaining 95.03% and 85.53% respectively in real-time simulations. With a net averaged anticipation time of 147.9 ms after 2.1 ms computation latency, this approach demonstrates that windows covering 74% of the gait cycle can synergize with deep learning to balance the inherent trade-off between extracting richer information and maintaining system responsiveness to changes in activity.

Human-in-the-Loop Myoelectric Pattern Recognition Control of an Arm-Support Robot to Improve Reaching in Stroke Survivors

The objective of this study was to assess the feasibility and efficacy of using real-time human-in-the-loop pattern recognition-based myoelectric control to control vertical support force or vertical position to improve reach in individuals with chronic stroke. This work attempts to move proven lab-based static arm support paradigms towards a controllable wearable device. A machine learning (linear discriminant analysis)-based myoelectric pattern recognition system based on movement intent as determined by real-time muscle activation was used to control incremental changes in either vertical position or vertical support force during a reach and retrieve task, with the goal of improving reaching function. Performance under real-time control of both options was compared to two unchanging static-support conditions (current gold standard) and a no-support condition. Both real-time control paradigms were successfully implemented and resulted in greater forward-reaching performance as demonstrated by increased elbow extension and horizontal shoulder adduction compared to no-support and was not different from the current gold standard static support paradigms. Muscle activation levels with real-time support were lower than the no-support condition and similar to those observed during the static support paradigms. Real-time detection of user intent was successful in controlling both vertical position and vertical support force and enabled greater reaching distance than without it demonstrating both its feasibility and efficacy albeit with some limitations.

Electromyographic typing gesture classification dataset for neurotechnological human-machine interfaces

Neurotechnological interfaces have the potential to create new forms of human-machine interactions, by allowing devices to interact directly with neurological signals instead of via intermediates such as keystrokes. Surface electromyography (sEMG) has been used extensively in myoelectric control systems, which use bioelectric activity recorded from muscles during contractions to classify actions. This technology has been used primarily for rehabilitation applications. In order to support the development of myoelectric interfaces for a broader range of human-machine interactions, we present an sEMG dataset obtained during key presses in a typing task. This fine-grained classification dataset consists of 16-channel bilateral sEMG recordings and key logs, collected from 19 individuals in two sessions on different days. We report baseline results on intra-session, inter-session and inter-subject evaluations. Our baseline results show that within-session accuracy is relatively high, even with simple learning models. However, the results on between-session and between-participant are much lower, showing that generalizing between sessions and individuals is an open challenge.

An assistive robot that enables people with amyotrophia to perform sequences of everyday activities

Mobile manipulation aids aim at enabling people with motor impairments to physically interact with their environment. To facilitate the operation of such systems, a variety of components, such as suitable user interfaces and intuitive control of the system, play a crucial role. In this article, we validate our highly integrated assistive robot EDAN, operated by an interface based on bioelectrical signals, combined with shared control and a whole-body coordination of the entire system, through a case study involving people with motor impairments to accomplish real-world activities. Three individuals with amyotrophia were able to perform a range of everyday tasks, including pouring a drink, opening and driving through a door, and opening a drawer. Rather than considering these tasks in isolation, our study focuses on the continuous execution of long sequences of realistic everyday tasks.

Comparing on-line continuous movement decoding with joints unconstrained and constrained based on a generic musculoskeletal model

Human-machine interface (HMI) has been extensively developed and applied in rehabilitation. However, the performance of amputees on continuous movement decoding was significantly decreased compared with that of able-bodied individuals. To explore the impact of the absence of joint movements on the performance of HMI in rehabilitation, a generic musculoskeletal model (MM) was employed in this study to evaluate and compare the performance of subjects completing a series of on-line tasks with the wrist and metacarpophalangeal (MCP) joints unconstrained and constrained. The performance of the generic MM has been demonstrated in previous studies. The electromyography (EMG) signals of four muscles were employed as inputs of the generic MM to realize the continuous movement decoding of wrist and MCP joints. Ten able-bodied subjects were recruited to perform the on-line tasks. The completion time, the number of overshoots, and the path efficiency of the tasks were taken as the indexes to quantify the subjects' performance. The muscle activation associated with the movement was analyzed. Across all tasks and subjects, the average values of the three indexes with the joints unconstrained were 7.7 s, 0.59, and 0.38, respectively, while those with the joints constrained were 17.86 s, 1.47, and 0.22, respectively. The results demonstrated that the subjects performed better with the wrist and MCP joints unconstrained than with those joints constrained in the on-line tasks, suggesting that the absence of joint movements can be a reason of the decreased performance of continuous movement decoding with HMIs. Meanwhile, it is revealed that the different performance on motion behaviors is caused by the absence of joint movements.

Examining the physical and psychological effects of combining multimodal feedback with continuous control in prosthetic hands

Myoelectric prosthetic hands are typically controlled to move between discrete positions and do not provide sensory feedback to the user. In this work, we present and evaluate a closed-loop, continuous myoelectric prosthetic hand controller, that can continuously control the position of multiple degrees of freedom of a prosthesis while rendering proprioceptive feedback to the user via a haptic feedback armband. Twenty-eight participants without and ten participants with upper limb difference (ULD) were recruited to holistically evaluate the physical and psychological effects of the controller via isolated control and sensory tasks, dexterity assessments, embodiment and task load questionnaires, and post-study interviews. The combination of proprioceptive feedback and continuous control enabled more accurate position and force modulation than without proprioceptive feedback, and restored blindfolded object identification ability to open-loop discrete controller levels. Dexterity assessment and embodiment questionnaire results revealed no significant physical performance or psychological embodiment differences between control types, with the exception of perceived sensation questions, which were significantly higher (p < 0.001) for closed-loop controllers. Key differences between participants with and without ULD were identified, including increasingly lower perceived body completeness and heterogeneity in frustration in participants with ULD, which can inform future development and rehabilitation.

A Survey of Few-Shot Learning for Biomedical Time Series

Advancements in wearable sensor technologies and the digitization of medical records have contributed to the unprecedented ubiquity of biomedical time series data. Data-driven models have tremendous potential to assist clinical diagnosis and improve patient care by improving long-term monitoring capabilities, facilitating early disease detection and intervention, as well as promoting personalized healthcare delivery. However, accessing extensively labeled datasets to train data-hungry deep learning models encounters many barriers, such as long-tail distribution of rare diseases, cost of annotation, privacy and security concerns, data-sharing regulations, and ethical considerations. An emerging approach to overcome the scarcity of labeled data is to augment AI methods with human-like capabilities to leverage past experiences to learn new tasks with limited examples, called few-shot learning. This survey provides a comprehensive review and comparison of few-shot learning methods for biomedical time series applications. The clinical benefits and limitations of such methods are discussed in relation to traditional data-driven approaches. This paper aims to provide insights into the current landscape of few-shot learning for biomedical time series and its implications for future research and applications.

Myoelectric pattern recognition with virtual reality and serious gaming improves upper limb function in chronic stroke: a single case experimental design study

BACKGROUND

Myoelectric pattern recognition (MPR) combines multiple surface electromyography channels with a machine learning algorithm to decode motor intention with an aim to enhance upper limb function after stroke. This study aims to determine the feasibility and preliminary effectiveness of a novel intervention combining MPR, virtual reality (VR), and serious gaming to improve upper limb function in people with chronic stroke.

METHODS

In this single case experimental A-B-A design study, six individuals with chronic stroke and moderate to severe upper limb impairment completed 18, 2 h sessions, 3 times a week. Repeated assessments were performed using the Fugl-Meyer Assessment of Upper Extremity (FMA-UE), Action Research Arm Test (ARAT), grip strength, and kinematics of the drinking task at baseline, during, and post intervention. The results were analyzed by using visual analysis and Tau-U statistics.

RESULTS

All participants improved upper limb function assessed by FMA-UE (Tau-U 0.72-1.0), and five out of six improved beyond the minimal clinical important difference (MCID). Four participants improved ARAT and grip strength scores (Tau-U 0.84-1.0), with one reaching the MCID for ARAT. Three out of four participants in the kinematic analysis achieved improvements beyond the MCID in movement time and smoothness, two with a Tau-U > 0.90, and two participants improved trunk displacement beyond the MCID (Tau-U 0.68). Most participants showed some deterioration in the follow-up phase.

CONCLUSIONS

MPR combined with VR and serious gaming is a feasible and promising intervention for improving upper limb function in people with chronic stroke.

TRIAL REGISTRATION

ClinicalTrials.gov, reference number NCT04154371.

EMG-based wake gestures eliminate false activations during out-of-set activities of daily living: an online myoelectric control study

Objective.While myoelectric control has been commercialized in prosthetics for decades, its adoption for more general human-machine interaction has been slow. Although high accuracies can be achieved across many gestures, current control approaches are prone to false activations in real-world conditions. This is because the same electromyogram (EMG) signals generated during the elicitation of gestures are also naturally activated when performing activities of daily living (ADLs), such as when driving to work or while typing on a keyboard. This can lead the myoelectric control system, which is trained on a closed set of gestures and thus unaware of the muscle activity associated with these ADLs, to be falsely activated, leading to erroneous inputs and user frustration.Approach.To overcome this problem, the concept of wake gestures, whereby users could switch between a dedicated control mode and a sleep mode by snapping their fingers, was explored. Using a simple dynamic time warping model, the real-world user-in-the-loop efficacy of wake gestures as a toggle for myoelectric interfaces was demonstrated through two online ubiquitous control tasks with varying levels of difficulty: (1) dismissing an alarm and (2) controlling a robot.Main results.During these online evaluations, the designed system ignored almost all (>99.9%) non-target EMG activity generated during a set of ADLs (i.e. walking, typing, writing, phone use, and driving), ignored all control gestures (i.e. wrist flexion, wrist extension, hand open, and hand close), and enabled reliable mode switching during intentional wake gesture elicitation. Additionally, questionnaires revealed that participants responded well to the use of wake gestures and generally preferred false negatives over false positives, providing valuable insights into the future design of these systems.Significance.These results highlight the real-world viability of wake gestures for enabling the intermittent use of myoelectric control, opening up new interaction possibilities for EMG-based inputs.

The Investigation of the Relationship Between Individual Pain Perception, Brain Electrical Activity, and Facial Expression Based on Combined EEG and Facial EMG Analysis

Purpose

Pain is a multidimensional, unpleasant emotional and sensory experience, and accurately assessing its intensity is crucial for effective management. However, individuals with cognitive impairments or language deficits may struggle to accurately report their pain. EEG provides insight into the neurological aspects of pain, while facial EMG captures the sensory and peripheral muscle responses. Our objective is to explore the relationship between individual pain perception, brain activity, and facial expressions through a combined analysis of EEG and facial EMG, aiming to provide an objective and multidimensional approach to pain assessment.

Methods

We investigated pain perception in response to electrical stimulation of the middle finger in 26 healthy subjects. The 32-channel EEG and 3-channel facial EMG signals were simultaneously recorded during a pain rating task. Group difference and correlation analysis were employed to investigate the relationship between individual pain perception, EEG, and facial EMG. The general linear model (GLM) was used for multidimensional pain assessment.

Results

The EEG analysis revealed that painful stimuli induced N2-P2 complex waveforms and gamma oscillations, with substantial variability in response to different stimuli. The facial EMG signals also demonstrated significant differences and variability correlated with subjective pain ratings. A combined analysis of EEG and facial EMG data using a general linear model indicated that both N2-P2 complex waveforms and the zygomatic muscle responses significantly contributed to pain assessment.

Conclusion

Facial EMG signals provide pain descriptions which are not sufficiently captured by EEG signals, and integrating both signals offers a more comprehensive understanding of pain perception. Our study underscores the potential of multimodal neurophysiological measurements in pain perception, offering a more comprehensive framework for evaluating pain.

Simultaneous and Proportional Control Based on an Enhanced Musculoskeletal Model

Recently, the musculoskeletal model (MM) has been widely studied for decoding movement intent from electromyography (EMG) signals. However, the decoding performance of the MM is impaired for the coordinated movements of multiple degrees of freedom (DoFs) due to the crosstalk between signals of multiple muscles. To address this problem, this study proposed an enhanced MM for 3-DoF motion prediction by taking the "divide and conquer" (DC) strategy and integrating the non-negative matrix factorization (NMF) algorithm, which is named as DC-NMF-MM. The control signals of wrist flexion/extension and MCP flexion/extension were obtained from four independent muscles, and the control signals of wrist pronation/supination were obtained from eight-channel surface EMG signals. Eight non-disabled subjects were recruited for offline and online experiment. For offline experiment, another two MMs were established and taken as the control groups for validation of the proposed DC-NMF-MM, including the MM totally taking the NMF algorithm (T-NMF-MM) and that partly taking the NMF algorithm (P-NMF-MM) for predicting the wrist pronation/supination only. The Pearson's correlation coefficient and the normalized root mean square error were employed to compare the prediction performance of three models. The results showed that the proposed method performs better than the other two models. Moreover, artificial neural network and linear regression model were established to compare with the proposed model and the results showed that DC-NMF-MM is more accurate in predicting joint Angle. For online experiment, a general 3-DOF musculoskeletal model based on DC-NMF-MM was established and the completion time, the number of overshoots, and the path efficiency were taken as evaluation indexes. The results further demonstrated the feasibility of the proposed method to achieve 3-DoF motion control. The proposed enhanced MM provides a prerequisite for the realization of clinical hand myoelectric control.

The extraction of neural strategies from the surface EMG: 2004-2024

This review follows two previous papers [Farina et al. Appl Physiol (1985) 96: 1486-1495, 2004; Farina et al. J Appl Physiol (1985) 117: 1215-1230, 2014] in which we reflected on the use of surface electromyography (EMG) in the study of the neural control of movement. This series of papers began with an analysis of the indirect approaches of EMG processing to infer the neural control strategies and then closely followed the progress in EMG technology. In this third paper, we focus on three main areas: surface EMG modeling; surface EMG processing, with an emphasis on decomposition; and interfacing applications of surface EMG recordings. We highlight the latest advances in EMG models that allow fast generation of simulated signals from realistic volume conductors, with applications ranging from validation of algorithms to identification of nonmeasurable parameters by inverse modeling. Surface EMG decomposition is currently an established state-of-the-art tool for physiological investigations of motor units. It is now possible to identify large samples of motor units, to track motor units over multiple sessions, to partially compensate for the nonstationarities in dynamic contractions, and to decompose signals in real time. The latter achievement has facilitated advances in myocontrol, by using the online decoded neural drive as a control signal, such as in the interfacing of prostheses. Looking back over the 20 yr since our first review, we conclude that the recording and analysis of surface EMG signals have seen breakthrough advances in this period. Although challenges in its application and interpretation remain, surface EMG is now a solid and unique tool for the study of the neural control of movement.

Decoding Gestures in Electromyography: Spatiotemporal Graph Neural Networks for Generalizable and Interpretable Classification

In recent years, significant strides in deep learning have propelled the advancement of electromyography (EMG)-based upper-limb gesture recognition systems, yielding notable successes across a spectrum of domains, including rehabilitation, orthopedics, robotics, and human-computer interaction. Despite these achievements, prevailing methodologies often overlook the intrinsic physical configurations and interconnectivity of multi-channel sensory inputs, resulting in a failure to adequately capture relational information embedded within the connections of deployed EMG sensor network topology. This oversight poses a significant challenge, impeding the extraction of crucial features from collaborative multi-channel EMG inputs and subsequently constraining model performance, generalizability, and interpretability. To address these limitations, we introduce novel graph structures meticulously crafted to encapsulate the spatial proximity of distributed EMG sensors and the temporal adjacency of EMG signals. Harnessing these tailored graph structures, we present Graph Convolution Network (GCN)-based classification models adept at effectively extracting and aggregating key features associated with various gestures. Our methodology exhibits remarkable efficacy, achieving state-of-the-art performance across five publicly available datasets, thus underscoring its prowess in gesture recognition tasks. Furthermore, our approach provides interpretable insights into muscular activation patterns, thereby reaffirming the practical effectiveness of our GCN model. Moreover, we show the effectiveness of our graph-based input structure and GCN-based classifier in maintaining high accuracy even with reduced sensor configurations, suggesting their potential for seamless integration into AI-powered rehabilitation strategies utilizing EMG-based gesture classification systems.

Surface electromyography evaluation for decoding hand motor intent in children with congenital upper limb deficiency

Children born with congenital upper limb absence exhibit consistent and distinguishable levels of biological control over their affected muscles, assessed through surface electromyography (sEMG). This represents a significant advancement in determining how these children might utilize sEMG-controlled dexterous prostheses. Despite this potential, the efficacy of employing conventional sEMG classification techniques for children born with upper limb absence is uncertain, as these techniques have been optimized for adults with acquired amputations. Tuning sEMG classification algorithms for this population is crucial for facilitating the successful translation of dexterous prostheses. To support this effort, we collected sEMG data from a cohort of N = 9 children with unilateral congenital below-elbow deficiency as they attempted 11 hand movements, including rest. Five classification algorithms were used to decode motor intent, tuned with features from the time, frequency, and time-frequency domains. We derived the congenital feature set (CFS) from the participant-specific tuned feature sets, which exhibited generalizability across our cohort. The CFS offline classification accuracy across participants was 73.8% ± 13.8% for the 11 hand movements and increased to 96.5% ± 6.6% when focusing on a reduced set of five movements. These results highlight the potential efficacy of individuals born with upper limb absence to control dexterous prostheses through sEMG interfaces.

Specifications and functional impact of a self-triggered grasp neuroprosthesis developed to restore prehension in hemiparetic post-stroke subjects

BACKGROUND

Stroke is the leading cause of acquired motor deficiencies in adults. Restoring prehension abilities is challenging for individuals who have not recovered active hand opening capacities after their rehabilitation. Self-triggered functional electrical stimulation applied to finger extensor muscles to restore grasping abilities in daily life is called grasp neuroprosthesis (GNP) and remains poorly accessible to the post-stroke population. Thus, we developed a GNP prototype with self-triggering control modalities adapted to the characteristics of the post-stroke population and assessed its impact on abilities.

METHODS

Through two clinical research protocols, 22 stroke participants used the GNP and its control modalities (EMG activity of a pre-defined muscle, IMU motion detection, foot switches and voice commands) for 3 to 5 sessions over a week. The NeuroPrehens software interpreted user commands through input signals from electromyographic, inertial, foot switches or microphone sensors to trigger an external electrical stimulator using two bipolar channels with surface electrodes. Users tested a panel of 9 control modalities, subjectively evaluated in ease-of-use and reliability with scores out of 10 and selected a preferred one before training with the GNP to perform functional unimanual standardized prehension tasks in a seated position. The responsiveness and functional impact of the GNP were assessed through a posteriori analysis of video recordings of these tasks across the two blinded evaluation multi-crossover N-of-1 randomized controlled trials.

RESULTS

Non-paretic foot triggering, whether from EMG or IMU, received the highest scores in both ease-of-use (median scores out of 10: EMG 10, IMU 9) and reliability (EMG 9, IMU 9) and were found viable and appreciated by users, like voice control and head lateral inclination modalities. The assessment of the system's general responsiveness combined with the control modalities latencies revealed median (95% confidence interval) durations between user intent and FES triggering of 333 ms (211 to 561), 217 ms (167 to 355) and 467 ms (147 to 728) for the IMU, EMG and voice control types of modalities, respectively. The functional improvement with the use of the GNP was significant in the two prehension tasks evaluated, with a median (95% confidence interval) improvement of 3 (- 1 to 5) points out of 5.

CONCLUSIONS

The GNP prototype and its control modalities were well suited to the post-stroke population in terms of self-triggering, responsiveness and restoration of functional grasping abilities. A wearable version of this device is being developed to improve prehension abilities at home.

TRIAL REGISTRATION

Both studies are registered on clinicaltrials.gov: NCT03946488, registered May 10, 2019 and NCT04804384, registered March 18, 2021.

Posture-invariant myoelectric control with self-calibrating random forests

Introduction

Myoelectric control systems translate different patterns of electromyographic (EMG) signals into the control commands of diverse human-machine interfaces via hand gesture recognition, enabling intuitive control of prosthesis and immersive interactions in the metaverse. The effect of arm position is a confounding factor leading to the variability of EMG characteristics. Developing a model with its characteristics and performance invariant across postures, could largely promote the translation of myoelectric control into real world practice.

Methods

Here we propose a self-calibrating random forest (RF) model which can (1) be pre-trained on data from many users, then one-shot calibrated on a new user and (2) self-calibrate in an unsupervised and autonomous way to adapt to varying arm positions.

Results

Analyses on data from 86 participants (66 for pre-training and 20 in real-time evaluation experiments) demonstrate the high generalisability of the proposed RF architecture to varying arm positions.

Discussion

Our work promotes the use of simple, explainable, efficient and parallelisable model for posture-invariant myoelectric control.

Wrist torque estimation by combining motor unit discharges with musculoskeletal model

OBJECTIVE

The application of electromyography (EMG) decomposition techniques in myoelectric control has gradually increased. However, most decomposition-based control schemes rely on machine learning, lacking interpretation of the biological mechanisms underlying movement generation and requiring large datasets for training. As neuromusculoskeletal modeling provides a promising alternative, this study proposes a decomposition-based musculoskeletal model for simultaneous and proportional myoelectric control.

METHODS

Sixteen able-bodied subjects participated in two experiments involving isometric wrist contractions in two degrees of freedom (DoF). High-density surface EMG signals and torques were recorded simultaneously. The EMG signals were decomposed into motor unit action potential trains (MUAPts). We proposed four clustering methods (two activation-based and two action potential-based) to group MUAPts, from which three neural features were extracted as neural excitations and input to the musculoskeletal model.

MAIN RESULTS

An activation-based clustering method with the twitch feature achieved a relatively high accuracy (R2 = 0.791 ± 0.101 and 0.622 ± 0.148 in the two experiments) with the highest smoothness (Roughness = 1.389 ± 0.211 and 1.140 ± 0.159).

CONCLUSION AND SIGNIFICANCE

The proposed MUAPt-based musculoskeletal model achieved promising accuracy in estimating continuous 2-DoF wrist torques, providing a novel approach for understanding the neuromechanical properties of multi-DoF movements and advancing the development of dexterous rehabilitation and robotic control.

Identification of electromyographic patterns of bradykinesia in patients with Parkinson's disease

Background

Parkinson's disease (PD) is a common neurodegenerative disease characterized by rest tremor, rigidity, and bradykinesia. Assessing the severity of these symptoms is typically done using the third part of the Movement Disorder Society-Sponsored Revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS III), relying on subjective evaluations by neurologists, which may lead to challenges in result interpretation. To address this issue, incorporation of surface electromyography (sEMG) in diagnostics.

Objectives

The aim of the study is to search for specific sEMG patterns that allow assessing the severity of bradykinesia.

Method

This case-control study involved 8 patients with PD at Hoehn & Yahr stages 2-3, and 7 healthy volunteers. sEMG was measured while the subjects performed the "finger tapping" and "hand movements" tests of the MDS-UPDRS III. The tests were conducted both before and after levodopa intake to identify patterns indicating changes in motor functions. During the tests, we observed the peak activity of the sEMG signal, reflecting the moments of activation of individual muscle groups involved in the implementation of the movement. Peak activity was characterized by the total number of maximum sEMG signal extrema and the distance between them. The data were compared with the healthy group.

Results

Peak activity increased after levodopa intake, indicating a reduction in bradykinesia. This feature partially correlates with clinicians' assessments and enhances the similarity of predictions by the MDS-UPDRS III scoring model to physician scores.

Conclusions

The results show the effectiveness of using sEMG and the number of peaks corresponding to the moments of muscle activation to assess bradykinesia.

Analysis of electrode locations on limb condition effect for myoelectric pattern recognition

BACKGROUND

Gesture recognition using surface electromyography (sEMG) has garnered significant attention due to its potential for intuitive and natural control in wearable human-machine interfaces. However, ensuring robustness remains essential and is currently the primary challenge for practical applications.

METHODS

This study investigates the impact of limb conditions and analyzes the influence of electrode placement. Both static and dynamic limb conditions were examined using electrodes positioned on the wrist, elbow, and the midpoint between them. Initially, we compared classification performance across various training conditions at these three electrode locations. Subsequently, a feature space analysis was conducted to quantify the effects of limb conditions. Finally, strategies for group training and feature selection were explored to mitigate these effects.

RESULTS

The results indicate that with the state-of-the-art method, classification performance at the wrist was comparable to that at the middle position, both of which outperformed the elbow, consistent with the findings from the feature space analysis. In inter-condition classification, training under dynamic limb conditions yielded better results than training under static conditions, especially at the positions covered by dynamic training. Additionally, fast and slow movement speeds produced similar performance outcomes. To mitigate the effects of limb conditions, adding more training conditions reduced classification errors; however, this reduction plateaued after four conditions, resulting in classification errors of 22.72%, 22.65%, and 26.58% for the wrist, middle, and elbow, respectively. Feature selection further improved classification performance, reducing errors to 19.98%, 19.75%, and 27.14% at the respective electrode locations, using three optimal features derived from single-condition training.

CONCLUSIONS

The study demonstrated that the impact of limb conditions was mitigated when electrodes were placed near the wrist. Dynamic limb condition training, combined with feature optimization, proved to be an effective strategy for reducing this effect. This work contributes to enhancing the robustness of myoelectric-controlled interfaces, thereby advancing the development of wearable intelligent devices.

Big data in myoelectric control: large multi-user models enable robust zero-shot EMG-based discrete gesture recognition

Myoelectric control, the use of electromyogram (EMG) signals generated during muscle contractions to control a system or device, is a promising input, enabling always-available control for emerging ubiquitous computing applications. However, its widespread use has historically been limited by the need for user-specific machine learning models because of behavioural and physiological differences between users. Leveraging the publicly available 612-user EMG-EPN612 dataset, this work dispels this notion, showing that true zero-shot cross-user myoelectric control is achievable without user-specific training. By taking a discrete approach to classification (i.e., recognizing the entire dynamic gesture as a single event), a classification accuracy of 93.0% for six gestures was achieved on a set of 306 unseen users, showing that big data approaches can enable robust cross-user myoelectric control. By organizing the results into a series of mini-studies, this work provides an in-depth analysis of discrete cross-user models to answer unknown questions and uncover new research directions. In particular, this work explores the number of participants required to build cross-user models, the impact of transfer learning for fine-tuning these models, and the effects of under-represented end-user demographics in the training data, among other issues. Additionally, in order to further evaluate the performance of the developed cross-user models, a completely new dataset was created (using the same recording device) that includes known covariate factors such as cross-day use and limb-position variability. The results show that the large data models can effectively generalize to new datasets and mitigate the impact of common confounding factors that have historically limited the adoption of EMG-based inputs.

Using Electrical Muscle Stimulation to Enhance Electrophysiological Performance of Agonist-Antagonist Myoneural Interface

The agonist-antagonist myoneural interface (AMI), a surgical method to reinnervate physiologically-relevant proprioceptive feedback for control of limb prostheses, has demonstrated the ability to provide natural afferent sensations for limb amputees when actuating their prostheses. Following AMI surgery, one potential challenge is atrophy of the disused muscles, which would weaken the reinnervation efficacy of AMI. It is well known that electrical muscle stimulus (EMS) can reduce muscle atrophy. In this study, we conducted an animal investigation to explore whether the EMS can significantly improve the electrophysiological performance of AMI. AMI surgery was performed in 14 rats, in which the distal tendons of bilateral solei donors were connected and positioned on the surface of the left biceps femoris. Subsequently, the left tibial nerve and the common peroneus nerve were sutured onto the ends of the connected donor solei. Two stimulation electrodes were affixed onto the ends of the donor solei for EMS delivery. The AMI rats were randomly divided into two groups. One group received the EMS treatment (designated as EMS_on) regularly for eight weeks and another received no EMS (designated as EMS_off). Two physiological parameters, nerve conduction velocity (NCV) and motor unit number, were derived from the electrically evoked compound action potential (CAP) signals to assess the electrophysiological performance of AMI. Our experimental results demonstrated that the reinnervated muscles of the EMS_on group generated higher CAP signals in comparison to the EMS_off group. Both NCV and motor unit number were significantly elevated in the EMS_on group. Moreover, the EMS_on group displayed statistically higher CAP signals on the indirectly activated proprioceptive afferents than the EMS_off group. These findings suggested that EMS treatment would be promising in enhancing the electrophysiological performance and facilitating the reinnervation process of AMI.

Improving Hand Gesture Recognition Robustness to Dynamic Posture Variations by Multimodal Deep Feature Fusion

Surface electromyography (sEMG), a human-machine interface for gesture recognition, has shown promising potential for decoding motor intentions, but a variety of nonideal factors restrict its practical application in assistive robots. In this paper, we summarized the current mainstream gesture recognition strategies and proposed a gesture recognition method based on multimodal canonical correlation analysis feature fusion classification (MCAFC) for a nonideal condition that occurs in daily life, i.e., posture variations. The deep features of the sEMG and acceleration signals were first extracted via convolutional neural networks. A canonical correlation analysis was subsequently performed to associate the deep features of the two modalities. The transformed features were utilized as inputs to a linear discriminant analysis classifier to recognize the corresponding gestures. Both offline and real-time experiments were conducted on eight non-disabled subjects. The experimental results indicated that MCAFC achieved an average classification accuracy, average motion completion rate, and average motion completion time of 93.44%, 94.05%, and 1.38 s, respectively, with multiple dynamic postures, indicating significantly better performance than that of comparable methods. The results demonstrate the feasibility and superiority of the proposed multimodal signal feature fusion method for gesture recognition with posture variations, providing a new scheme for myoelectric control.

Surface Electromyography-Based Recognition of Electronic Taste Sensations

Taste sensation recognition is a core for taste-related queries. Most prior research has been devoted to recognizing the basic taste sensations using the Brain-Computer Interface (BCI), which includes EEG, MEG, EMG, and fMRI. This research aims to recognize electronic taste (E-Taste) sensations based on surface electromyography (sEMG). Silver electrodes with platinum plating of the E-Taste device were placed on the tongue's tip to stimulate various tastes and flavors. In contrast, the electrodes of the sEMG were placed on facial muscles to collect the data. The dataset was organized and preprocessed, and a random forest classifier was applied, giving a five-fold accuracy of 70.43%. The random forest classifier was used on each participant dataset individually and in groups, providing the highest accuracy of 84.79% for a single participant. Moreover, various feature combinations were extracted and acquired 72.56% accuracy after extracting eight features. For a future perspective, this research offers guidance for electronic taste recognition based on sEMG.

Excitation of the abdominal ganglion affects the electrophysiological activity of indirect flight muscles of the honeybee Apis mellifera

Our understanding of the nervous tissues that affect the wing flapping of insects mainly focuses on the brain, but wing flapping is a rhythmic movement related to the central pattern generator in the ventral nerve cord. To verify whether the neural activity of the abdominal ganglion of the honeybee (Apis mellifera) affects the flapping-wing flight, we profiled the response characteristics of indirect flight muscles to abdominal ganglion excitation. Strikingly, a change in the neural activity of ganglion 3 or ganglion 4 has a stronger effect on the electrophysiological activity of indirect flight muscles than that of ganglion 5. The electrophysiological activity of vertical indirect flight muscles is affected more by the change in neural activity of the abdominal ganglion than that of lateral indirect flight muscles. Moreover, the change in neural activity of the abdominal ganglion mainly causes the change in the muscular activity of indirect wing muscles, but the activity patterns change relatively little and there is little change in the complicated details. This work improves our understanding of the neuroregulatory mechanisms associated with the flapping-wing flight of honeybees.

Modeling the Intent to Interact With VR Using Physiological Features

OBJECTIVE

Mixed-Reality (XR) technologies promise a user experience (UX) that rivals the interactive experience with the real-world. The key facilitators in the design of such a natural UX are that the interaction has zero lag and that users experience no excess mental load. This is difficult to achieve due to technical constraints such as motion-to-photon latency as well as false-positives during gesture-based interaction.

METHODS

In this paper, we explored the use of physiological features to model the user's intent to interact with a virtual reality (VR) environment. Accurate predictions about when users want to express an interaction intent could overcome the limitations of an interactive device that lags behind the intention of a user. We computed time-domain features from electroencephalography (EEG) and electromyography (EMG) recordings during a grab-and-drop task in VR and cross-validated a Linear Discriminant Analysis (LDA) for three different combinations of (1) EEG, (2) EMG and (3) EEG-EMG features.

RESULTS & CONCLUSION

We found the classifiers to detect the presence of a pre-movement state from background idle activity reflecting the users' intent to interact with the virtual objects (EEG: 62 % ± 10 %, EMG: 72 % ± 9 %, EEG-EMG: 69 % ± 10 %) above simulated chance level. The features leveraged in our classification scheme have a low computational cost and are especially useful for fast decoding of users' mental states. Our work is a further step towards a useful classification of users' intent to interact, as a high temporal resolution and speed of detection is crucial. This facilitates natural experiences through zero-lag adaptive interfaces.

sEMG-Driven Hand Dynamics Estimation With Incremental Online Learning on a Parallel Ultra-Low-Power Microcontroller

Surface electromyography (sEMG) is a State-of-the-Art (SoA) sensing modality for non-invasive human-machine interfaces for consumer, industrial, and rehabilitation use cases. The main limitation of the current sEMG-driven control policies is the sEMG's inherent variability, especially cross-session due to sensor repositioning; this limits the generalization of the Machine/Deep Learning (ML/DL) in charge of the signal-to-command mapping. The other hot front on the ML/DL side of sEMG-driven control is the shift from the classification of fixed hand positions to the regression of hand kinematics and dynamics, promising a more versatile and fluid control. We present an incremental online-training strategy for sEMG-based estimation of simultaneous multi-finger forces, using a small Temporal Convolutional Network suitable for embedded learning-on-device. We validate our method on the HYSER dataset, cross-day. Our incremental online training reaches a cross-day Mean Absolute Error (MAE) of (9.58 ± 3.89)% of the Maximum Voluntary Contraction on HYSER's RANDOM dataset of improvised, non-predefined force sequences, which is the most challenging and closest to real scenarios. This MAE is on par with an accuracy-oriented, non-embeddable offline training exploiting more epochs. Further, we demonstrate that our online training approach can be deployed on the GAP9 ultra-low power microcontroller, obtaining a latency of 1.49 ms and an energy draw of just 40.4 uJ per forward-backward-update step. These results show that our solution fits the requirements for accurate and real-time incremental training-on-device.

Integrative approach to pedobarography and pelvis-trunk motion for knee osteoarthritis detection and exploration of non-radiographic rehabilitation monitoring

Background

Osteoarthritis (OA) is a highly prevalent global musculoskeletal disorder, and knee OA (KOA) accounts for four-fifths of the cases worldwide. It is a degenerative disorder that greatly affects the quality of life. Thus, it is managed through different methods, such as weight loss, physical therapy, and knee arthroplasty. Physical therapy aims to strengthen the knee periarticular muscles to improve joint stability.

Methods

Pedobarographic data and pelvis and trunk motion of 56 adults are recorded. Among them, 28 subjects were healthy, and 28 subjects were suffering from varying degrees of KOA. Age, sex, BMI, and the recorded variables are used together to identify subjects with KOA using machine learning (ML) models, namely, logistic regression, SVM, decision tree, and random forest. Surface electromyography (sEMG) signals are also recorded bilaterally from two muscles, the rectus femoris and biceps femoris caput longus, bilaterally during various activities for two healthy and six KOA subjects. Cluster analysis is then performed using the principal components obtained from time-series features, frequency features, and time-frequency features.

Results

KOA is successfully identified using the pedobarographic data and the pelvis and trunk motion with the highest accuracy and sensitivity of 89.3% and 85.7%, respectively, using a decision tree classifier. In addition, sEMG data have been successfully used to cluster healthy subjects from KOA subjects, with wavelet analysis features providing the best performance for the standing activity under different conditions.

Conclusion

KOA is detected using gait variables not directly related to the knee, such as pedobarographic measurements and pelvis and trunk motion captured by pedobarography mats and wearable sensors, respectively. KOA subjects are also distinguished from healthy individuals through clustering analysis using sEMG data from knee periarticular muscles during walking and standing. Gait data and sEMG complement each other, aiding in KOA identification and rehabilitation monitoring. It is important because wearable sensors simplify data collection, require minimal sample preparation, and offer a non-radiographic, safe method suitable for both laboratory and real-world scenarios. The decision tree classifier, trained with stratified k-fold cross validation (SKCV) data, is observed to be the best for KOA identification using gait data.

Unravelling Influence Factors in Pattern Recognition Myoelectric Control Systems: The Impact of Limb Positions and Electrode Shifts

Pattern recognition (PR)-based myoelectric control systems can naturally provide multifunctional and intuitive control of upper limb prostheses and restore lost limb function, but understanding their robustness remains an open scientific question. This study investigates how limb positions and electrode shifts-two factors that have been suggested to cause classification deterioration-affect classifiers' performance by quantifying changes in the class distribution using each factor as a class and computing the repeatability and modified separability indices. Ten intact-limb participants took part in the study. Linear discriminant analysis (LDA) was used as the classifier. The results confirmed previous studies that limb positions and electrode shifts deteriorate classification performance (14-21% decrease) with no difference between factors (p > 0.05). When considering limb positions and electrode shifts as classes, we could classify them with an accuracy of 96.13 ± 1.44% and 65.40 ± 8.23% for single and all motions, respectively. Testing on five amputees corroborated the above findings. We have demonstrated that each factor introduces changes in the feature space that are statistically new class instances. Thus, the feature space contains two statistically classifiable clusters when the same motion is collected in two different limb positions or electrode shifts. Our results are a step forward in understanding PR schemes' challenges for myoelectric control of prostheses and further validation needs be conducted on more amputee-related datasets.

Wrist EMG-based Gestures Recognition for Finger and Wrist Motions

Gesture recognition is a relatively natural humanmachine interface (HMI). Electromyography (EMG) based gesture recognition methods have been extensively investigated in upper limb prostheses as a special HMI, in which EMG sensors are mostly mounted on the proximal part of the forearm. However, for more general applications beyond upper-limb prosthetics, the wrist may be a more suitable position for more intuitive HMIs, in which independent finger movements in addition to wrist gestures can be realized. In this study, we propose to investigate the recognition performance for gestures of the index finger using wrist EMG. All DOFs of the metacarpophalangeal joint, including static and dynamic gestures with directions, were investigated. Forearm EMG and conventional wrist motions were used as controls for wrist EMG and finger motions, respectively. The frequency division technique (FDT) was first adopted for feature extraction of wrist EMG signals. Finally, three combinations of algorithm and feature were applied to gesture recognition. Results showed that linear discriminate analysis (LDA) and FDT using wrist EMG had a mean classification accuracy of 79% and 89% for static finger and wrist gestures, respectively, and for forearm EMG, the corresponding values were 73% and 90%. A potential biomedical application is to assist patients with index finger disability with the unobtrusive wrist-worn band.

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

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

A distributed, high-channel-count, implanted bidirectional system for restoration of somatosensation and myoelectric control

Objective. We intend to chronically restore somatosensation and provide high-fidelity myoelectric control for those with limb loss via a novel, distributed, high-channel-count, implanted system.Approach.We have developed the implanted Somatosensory Electrical Neurostimulation and Sensing (iSens®) system to support peripheral nerve stimulation through up to 64, 96, or 128 electrode contacts with myoelectric recording from 16, 8, or 0 bipolar sites, respectively. The rechargeable central device has Bluetooth® wireless telemetry to communicate to external devices and wired connections for up to four implanted satellite stimulation or recording devices. We characterized the stimulation, recording, battery runtime, and wireless performance and completed safety testing to support its use in human trials.Results.The stimulator operates as expected across a range of parameters and can schedule multiple asynchronous, interleaved pulse trains subject to total charge delivery limits. Recorded signals in saline show negligible stimulus artifact when 10 cm from a 1 mA stimulating source. The wireless telemetry range exceeds 1 m (direction and orientation dependent) in a saline torso phantom. The bandwidth supports 100 Hz bidirectional update rates of stimulation commands and data features or streaming select full bandwidth myoelectric signals. Preliminary first-in-human data validates the bench testing result.Significance.We developed, tested, and clinically implemented an advanced, modular, fully implanted peripheral stimulation and sensing system for somatosensory restoration and myoelectric control. The modularity in electrode type and number, including distributed sensing and stimulation, supports a wide variety of applications; iSens® is a flexible platform to bring peripheral neuromodulation applications to clinical reality. ClinicalTrials.gov ID NCT04430218.

Topology of surface electromyogram signals: hand gesture decoding on Riemannian manifolds

Objective. Decoding gestures from the upper limb using noninvasive surface electromyogram (sEMG) signals is of keen interest for the rehabilitation of amputees, artificial supernumerary limb augmentation, gestural control of computers, and virtual/augmented realities. We show that sEMG signals recorded across an array of sensor electrodes in multiple spatial locations around the forearm evince a rich geometric pattern of global motor unit (MU) activity that can be leveraged to distinguish different hand gestures.Approach. We demonstrate a simple technique to analyze spatial patterns of muscle MU activity within a temporal window and show that distinct gestures can be classified in both supervised and unsupervised manners. Specifically, we construct symmetric positive definite covariance matrices to represent the spatial distribution of MU activity in a time window of interest, calculated as pairwise covariance of electrical signals measured across different electrodes.Main results. This allows us to understand and manipulate multivariate sEMG timeseries on a more natural subspace-the Riemannian manifold. Furthermore, it directly addresses signal variability across individuals and sessions, which remains a major challenge in the field. sEMG signals measured at a single electrode lack contextual information such as how various anatomical and physiological factors influence the signals and how their combined effect alters the evident interaction among neighboring muscles.Significance. As we show here, analyzing spatial patterns using covariance matrices on Riemannian manifolds allows us to robustly model complex interactions across spatially distributed MUs and provides a flexible and transparent framework to quantify differences in sEMG signals across individuals. The proposed method is novel in the study of sEMG signals and its performance exceeds the current benchmarks while being computationally efficient.

DistaNet: grasp-specific distance biofeedback promotes the retention of myoelectric skills

Objective.An active myoelectric interface responds to the user's muscle signals to enable movements. Machine learning can decode user intentions from myoelectric signals. However, machine learning-based interface control lacks continuous, intuitive feedback about task performance, needed to facilitate the acquisition and retention of myoelectric control skills.Approach.We propose DistaNet as a neural network-based framework that extracts smooth, continuous, and low-dimensional signatures of the hand grasps from multi-channel myoelectric signals and provides grasp-specific biofeedback to the users.Main results.Experimental results show its effectiveness in decoding user gestures and providing biofeedback, helping users retain the acquired motor skills.Significance.We demonstrates myoelectric skill retention in a pattern recognition setting for the first time.

Dual Stream Long Short-Term Memory Feature Fusion Classifier for Surface Electromyography Gesture Recognition

Gesture recognition using electromyography (EMG) signals has prevailed recently in the field of human-computer interactions for controlling intelligent prosthetics. Currently, machine learning and deep learning are the two most commonly employed methods for classifying hand gestures. Despite traditional machine learning methods already achieving impressive performance, it is still a huge amount of work to carry out feature extraction manually. The existing deep learning methods utilize complex neural network architectures to achieve higher accuracy, which will suffer from overfitting, insufficient adaptability, and low recognition accuracy. To improve the existing phenomenon, a novel lightweight model named dual stream LSTM feature fusion classifier is proposed based on the concatenation of five time-domain features of EMG signals and raw data, which are both processed with one-dimensional convolutional neural networks and LSTM layers to carry out the classification. The proposed method can effectively capture global features of EMG signals using a simple architecture, which means less computational cost. An experiment is conducted on a public DB1 dataset with 52 gestures, and each of the 27 subjects repeats every gesture 10 times. The accuracy rate achieved by the model is 89.66%, which is comparable to that achieved by more complex deep learning neural networks, and the inference time for each gesture is 87.6 ms, which can also be implied in a real-time control system. The proposed model is validated using a subject-wise experiment on 10 out of the 40 subjects in the DB2 dataset, achieving a mean accuracy of 91.74%. This is illustrated by its ability to fuse time-domain features and raw data to extract more effective information from the sEMG signal and select an appropriate, efficient, lightweight network to enhance the recognition results.

S-WD-EEMD: A hybrid framework for imbalanced sEMG signal analysis in diagnosis of human knee abnormality

The diagnosis of human knee abnormalities using the surface electromyography (sEMG) signal obtained from lower limb muscles with machine learning is a major problem due to the noisy nature of the sEMG signal and the imbalance in data corresponding to healthy and knee abnormal subjects. To address this challenge, a combination of wavelet decomposition (WD) with ensemble empirical mode decomposition (EEMD) and the Synthetic Minority Oversampling Technique (S-WD-EEMD) is proposed. In this study, a hybrid WD-EEMD is considered for the minimization of noises produced in the sEMG signal during the collection, while the Synthetic Minority Oversampling Technique (SMOTE) is considered to balance the data by increasing the minority class samples during the training of machine learning techniques. The findings indicate that the hybrid WD-EEMD with SMOTE oversampling technique enhances the efficacy of the examined classifiers when employed on the imbalanced sEMG data. The F-Score of the Extra Tree Classifier, when utilizing WD-EEMD signal processing with SMOTE oversampling, is 98.4%, whereas, without the SMOTE oversampling technique, it is 95.1%.

Embedded Machine Learning System for Muscle Patterns Detection in a Patient with Shoulder Disarticulation

The integration of artificial intelligence (AI) models in the classification of electromyographic (EMG) signals represents a significant advancement in the design of control systems for prostheses. This study explores the development of a portable system that classifies the electrical activity of three shoulder muscles in real time for actuator control, marking a milestone in the autonomy of prosthetic devices. Utilizing low-power microcontrollers, the system ensures continuous EMG signal recording, enhancing user mobility. Focusing on a case study-a 42-year-old man with left shoulder disarticulation-EMG activity was recorded over two days using a specifically designed electronic board. Data processing was performed using the Edge Impulse platform, renowned for its effectiveness in implementing AI on edge devices. The first day was dedicated to a training session with 150 repetitions spread across 30 trials and three different movements. Based on these data, the second day tested the AI model's ability to classify EMG signals in new movement executions in real time. The results demonstrate the potential of portable AI-based systems for prosthetic control, offering accurate and swift EMG signal classification that enhances prosthetic user functionality and experience. This study not only underscores the feasibility of real-time EMG signal classification but also paves the way for future research on practical applications and improvements in the quality of life for prosthetic users.

Intelligent Human-Computer Interaction: Combined Wrist and Forearm Myoelectric Signals for Handwriting Recognition

Recent studies have highlighted the possibility of using surface electromyographic (EMG) signals to develop human-computer interfaces that are also able to recognize complex motor tasks involving the hand as the handwriting of digits. However, the automatic recognition of words from EMG information has not yet been studied. The aim of this study is to investigate the feasibility of using combined forearm and wrist EMG probes for solving the handwriting recognition problem of 30 words with consolidated machine-learning techniques and aggregating state-of-the-art features extracted in the time and frequency domains. Six healthy subjects, three females and three males aged between 25 and 40 years, were recruited for the study. Two tests in pattern recognition were conducted to assess the possibility of classifying fine hand movements through EMG signals. The first test was designed to assess the feasibility of using consolidated myoelectric control technology with shallow machine-learning methods in the field of handwriting detection. The second test was implemented to assess if specific feature extraction schemes can guarantee high performances with limited complexity of the processing pipeline. Among support vector machine, linear discriminant analysis, and K-nearest neighbours (KNN), the last one showed the best classification performances in the 30-word classification problem, with a mean accuracy of 95% and 85% when using all the features and a specific feature set known as TDAR, respectively. The obtained results confirmed the validity of using combined wrist and forearm EMG data for intelligent handwriting recognition through pattern recognition approaches in real scenarios.

Context-informed incremental learning improves both the performance and resilience of myoelectric control

Despite its rich history of success in controlling powered prostheses and emerging commercial interests in ubiquitous computing, myoelectric control continues to suffer from a lack of robustness. In particular, EMG-based systems often degrade over prolonged use resulting in tedious recalibration sessions, user frustration, and device abandonment. Unsupervised adaptation is one proposed solution that updates a model's parameters over time based on its own predictions during real-time use to maintain robustness without requiring additional user input or dedicated recalibration. However, these strategies can actually accelerate performance deterioration when they begin to classify (and thus adapt) incorrectly, defeating their own purpose. To overcome these limitations, we propose a novel adaptive learning strategy, Context-Informed Incremental Learning (CIIL), that leverages in situ context to better inform the prediction of pseudo-labels. In this work, we evaluate these CIIL strategies in an online target acquisition task for two use cases: (1) when there is a lack of training data and (2) when a drastic and enduring alteration in the input space has occurred. A total of 32 participants were evaluated across the two experiments. The results show that the CIIL strategies significantly outperform the current state-of-the-art unsupervised high-confidence adaptation and outperform models trained with the conventional screen-guided training approach, even after a 45-degree electrode shift (p < 0.05). Consequently, CIIL has substantial implications for the future of myoelectric control, potentially reducing the training burden while bolstering model robustness, and leading to improved real-time control.

The essential role of climate policy uncertainty in carbon emissions: a fresh insight

This study assesses the influence of climate policy uncertainty (CPU) on carbon emissions (CE) against the backdrop of economic policy uncertainty (EPU) in the US. The wavelet analysis provides a comprehensive understanding of correlations in the time and frequency domains. The results demonstrate a significant correlation between CPU and CE, which varies across different time periods and frequencies. In the time domain, the results indicate that the CPU and CE move together during certain subperiods. Moreover, there are observable comovements in the frequency domain, particularly in the short to medium range. However, the correlation becomes stronger in the short term when there is no EPU, suggesting a closer interaction between CPU and CE. Therefore, it is crucial for governments to prioritize improving the clarity, credibility, and consistency of climate policies. They should also consider potential economic shocks when designing these policies.

Decoding of unimanual and bimanual reach-and-grasp actions from EMG and IMU signals in persons with cervical spinal cord injury

Objective. Chronic motor impairments of arms and hands as the consequence of a cervical spinal cord injury (SCI) have a tremendous impact on activities of daily life. A considerable number of people however retain minimal voluntary motor control in the paralyzed parts of the upper limbs that are measurable by electromyography (EMG) and inertial measurement units (IMUs). An integration into human-machine interfaces (HMIs) holds promise for reliable grasp intent detection and intuitive assistive device control.Approach. We used a multimodal HMI incorporating EMG and IMU data to decode reach-and-grasp movements of groups of persons with cervical SCI (n = 4) and without (control, n = 13). A post-hoc evaluation of control group data aimed to identify optimal parameters for online, co-adaptive closed-loop HMI sessions with persons with cervical SCI. We compared the performance of real-time, Random Forest-based movement versus rest (2 classes) and grasp type predictors (3 classes) with respect to their co-adaptation and evaluated the underlying feature importance maps.Main results. Our multimodal approach enabled grasp decoding significantly better than EMG or IMU data alone (p<0.05). We found the 0.25 s directly prior to the first touch of an object to hold the most discriminative information. Our HMIs correctly predicted 79.3 ± STD 7.4 (102.7 ± STD 2.3 control group) out of 105 trials with grand average movement vs. rest prediction accuracies above 99.64% (100% sensitivity) and grasp prediction accuracies of 75.39 ± STD 13.77% (97.66 ± STD 5.48% control group). Co-adaption led to higher prediction accuracies with time, and we could identify adaptions in feature importances unique to each participant with cervical SCI.Significance. Our findings foster the development of multimodal and adaptive HMIs to allow persons with cervical SCI the intuitive control of assistive devices to improve personal independence.

Transferable non-invasive modal fusion-transformer (NIMFT) for end-to-end hand gesture recognition

Objective.Recent studies have shown that integrating inertial measurement unit (IMU) signals with surface electromyographic (sEMG) can greatly improve hand gesture recognition (HGR) performance in applications such as prosthetic control and rehabilitation training. However, current deep learning models for multimodal HGR encounter difficulties in invasive modal fusion, complex feature extraction from heterogeneous signals, and limited inter-subject model generalization. To address these challenges, this study aims to develop an end-to-end and inter-subject transferable model that utilizes non-invasively fused sEMG and acceleration (ACC) data.Approach.The proposed non-invasive modal fusion-transformer (NIMFT) model utilizes 1D-convolutional neural networks-based patch embedding for local information extraction and employs a multi-head cross-attention (MCA) mechanism to non-invasively integrate sEMG and ACC signals, stabilizing the variability induced by sEMG. The proposed architecture undergoes detailed ablation studies after hyperparameter tuning. Transfer learning is employed by fine-tuning a pre-trained model on new subject and a comparative analysis is performed between the fine-tuning and subject-specific model. Additionally, the performance of NIMFT is compared to state-of-the-art fusion models.Main results.The NIMFT model achieved recognition accuracies of 93.91%, 91.02%, and 95.56% on the three action sets in the Ninapro DB2 dataset. The proposed embedding method and MCA outperformed the traditional invasive modal fusion transformer by 2.01% (embedding) and 1.23% (fusion), respectively. In comparison to subject-specific models, the fine-tuning model exhibited the highest average accuracy improvement of 2.26%, achieving a final accuracy of 96.13%. Moreover, the NIMFT model demonstrated superiority in terms of accuracy, recall, precision, and F1-score compared to the latest modal fusion models with similar model scale.Significance.The NIMFT is a novel end-to-end HGR model, utilizes a non-invasive MCA mechanism to integrate long-range intermodal information effectively. Compared to recent modal fusion models, it demonstrates superior performance in inter-subject experiments and offers higher training efficiency and accuracy levels through transfer learning than subject-specific approaches.