Prediction of muscle activity during loaded movements of the upper limb

The concepts of muscle activity generation driven by upper limb kinematics

BACKGROUND

The underlying motivation of this work is to demonstrate that artificial muscle activity of known and unknown motion can be generated based on motion parameters, such as angular position, acceleration, and velocity of each joint (or the end-effector instead), which are similarly represented in our brains. This model is motivated by the known motion planning process in the central nervous system. That process incorporates the current body state from sensory systems and previous experiences, which might be represented as pre-learned inverse dynamics that generate associated muscle activity.

METHODS

We develop a novel approach utilizing recurrent neural networks that are able to predict muscle activity of the upper limbs associated with complex 3D human arm motions. Therefore, motion parameters such as joint angle, velocity, acceleration, hand position, and orientation, serve as input for the models. In addition, these models are trained on multiple subjects (n=5 including , 3 male in the age of 26±2 years) and thus can generalize across individuals. In particular, we distinguish between a general model that has been trained on several subjects, a subject-specific model, and a specific fine-tuned model using a transfer learning approach to adapt the model to a new subject. Estimators such as mean square error MSE, correlation coefficient r, and coefficient of determination R² are used to evaluate the goodness of fit. We additionally assess performance by developing a new score called the zero-line score. The present approach was compared with multiple other architectures.

RESULTS

The presented approach predicts the muscle activity for previously through different subjects with remarkable high precision and generalizing nicely for new motions that have not been trained before. In an exhausting comparison, our recurrent network outperformed all other architectures. In addition, the high inter-subject variation of the recorded muscle activity was successfully handled using a transfer learning approach, resulting in a good fit for the muscle activity for a new subject.

CONCLUSIONS

The ability of this approach to efficiently predict muscle activity contributes to the fundamental understanding of motion control. Furthermore, this approach has great potential for use in rehabilitation contexts, both as a therapeutic approach and as an assistive device. The predicted muscle activity can be utilized to guide functional electrical stimulation, allowing specific muscles to be targeted and potentially improving overall rehabilitation outcomes.

Hand Posture and Force Estimation Using Surface Electromyography and an Artificial Neural Network

OBJECTIVE

The purpose of this study was to develop an approach to predict hand posture (pinch versus grip) and grasp force using forearm surface electromyography (sEMG) and artificial neural networks (ANNs) during tasks that varied repetition rate and duty cycle.

BACKGROUND

Prior studies have used electromyography with machine learning models to predict grip force but relatively few studies have assessed whether both hand posture and force can be predicted, particularly at varying levels of duty cycle and repetition rate.

METHOD

Fourteen individuals participated in this experiment. sEMG data for five forearm muscles and force output data were collected. Calibration data (25, 50, 75, 100% of maximum voluntary contraction (MVC)) were used to train ANN models to predict hand posture (pinch versus grip) and force magnitude while performing tasks that varied load, repetition rate, and duty cycle.

RESULTS

Across all participants, overall hand posture prediction accuracy was 79% (0.79 ± .08), whereas overall hand force prediction accuracy was 73% (0.73 ± .09). Accuracy ranged between 0.65 and 0.93 based on varying repetition rate and duty cycle.

CONCLUSION

Hand posture and force prediction were possible using sEMG and ANNs, though there were important differences in the accuracy of predictions based on task characteristics including duty cycle and repetition rate.

APPLICATION

The results of this study could be applied to the development of a dosimeter used for distal upper extremity biomechanical exposure measurement, risk assessment, job (re)design, and return to work programs.

Two-stage classification of electromyogram signals from hand grasps in the transverse plane

This paper presents a two-stage classification to resolve the effect of arm position changes on electromyogram (EMG) classification for hand grasps in the transverse plane. The proposed method combines the EMG signals with the signals from an inertial measurement unit in both the position and motion classification stages. To improve accuracy, we incorporate EMG data from the upper arm and shoulder with the forearm EMG signals. When evaluated on the five alternative object grasps placed on the nine positions, the proposed technique yields an average total classification error of 0.9%, which is a substantial improvement over the single-stage classification (4.3%).

Restoration of complex movement in the paralyzed upper limb

OBJECTIVE

Functional electrical stimulation (FES) involves artificial activation of skeletal muscles to reinstate motor function in paralyzed individuals. While FES applied to the upper limb has improved the ability of tetraplegics to perform activities of daily living, there are key shortcomings impeding its widespread use. One major limitation is that the range of motor behaviors that can be generated is restricted to a small set of simple, preprogrammed movements. This limitation stems from the substantial difficulty in determining the patterns of stimulation across many muscles required to produce more complex movements. Therefore, the objective of this study was to use machine learning to flexibly identify patterns of muscle stimulation needed to evoke a wide array of multi-joint arm movements.

APPROACH

Arm kinematics and electromyographic activity from 29 muscles were recorded while a 'trainer' monkey made an extensive range of arm movements. Those data were used to train an artificial neural network that predicted patterns of muscle activity associated with a new set of movements. Those patterns were converted into trains of stimulus pulses that were delivered to upper limb muscles in two other temporarily paralyzed monkeys.

RESULTS

Machine-learning based prediction of EMG was good for within-subject predictions but appreciably poorer for across-subject predictions. Evoked responses matched the desired movements with good fidelity only in some cases. Means to mitigate errors associated with FES-evoked movements are discussed.

SIGNIFICANCE

Because the range of movements that can be produced with our approach is virtually unlimited, this system could greatly expand the repertoire of movements available to individuals with high level paralysis.

InverseMuscleNET: Alternative Machine Learning Solution to Static Optimization and Inverse Muscle Modeling

InverseMuscleNET, a machine learning model, is proposed as an alternative to static optimization for resolving the redundancy issue in inverse muscle models. A recurrent neural network (RNN) was optimally configured, trained, and tested to estimate the pattern of muscle activation signals. Five biomechanical variables (joint angle, joint velocity, joint acceleration, joint torque, and activation torque) were used as inputs to the RNN. A set of surface electromyography (EMG) signals, experimentally measured around the shoulder joint for flexion/extension, were used to train and validate the RNN model. The obtained machine learning model yields a normalized regression in the range of 88-91% between experimental data and estimated muscle activation. A sequential backward selection algorithm was used as a sensitivity analysis to discover the less dominant inputs. The order of most essential signals to least dominant ones was as follows: joint angle, activation torque, joint torque, joint velocity, and joint acceleration. The RNN model required 0.06 s of the previous biomechanical input signals and 0.01 s of the predicted feedback EMG signals, demonstrating the dynamic temporal relationships of the muscle activation profiles. The proposed approach permits a fast and direct estimation ability instead of iterative solutions for the inverse muscle model. It raises the possibility of integrating such a model in a real-time device for functional rehabilitation and sports evaluation devices with real-time estimation and tracking. This method provides clinicians with a means of estimating EMG activity without an invasive electrode setup.

Resting Tendon Cross-Sectional Area Underestimates Biceps Brachii Tendon Stress: Importance of Measuring During a Contraction

Force produced by the muscle during contraction is applied to the tendon and distributed through the cross-sectional area (CSA) of the tendon. This ratio of force to the tendon CSA is quantified as the tendon mechanical property of stress. Stress is traditionally calculated using the resting tendon CSA; however, this does not take into account the reductions in the CSA resulting from tendon elongation during the contraction. It is unknown if calculating the tendon stress using instantaneous CSA during a contraction significantly increases the values of in vivo distal biceps brachii (BB) tendon stress in humans compared to stress calculated with the resting CSA. Nine young (22 ± 1 years) and nine old (76 ± 4 years) males, and eight young females (21 ± 1 years) performed submaximal isometric elbow flexion tracking tasks at force levels ranging from 2.5 to 80% maximal voluntary contraction (MVC). The distal BB tendon CSA was recorded on ultrasound at rest and during the submaximal tracking tasks (instantaneous). Tendon stress was calculated as the ratio of tendon force during contraction to CSA using the resting and instantaneous measures of CSA, and statistically evaluated with multi-level modeling (MLM) and Johnson–Neyman regions of significance tests to determine the specific force levels above which the differences between calculation methods and groups became statistically significant. The tendon CSA was greatest at rest and decreased as the force level increased (p < 0.001), and was largest in young males (23.0 ± 2.90 mm²) followed by old males (20.87 ± 2.0 mm²) and young females (17.08 ± 1.54 mm²) (p < 0.001) at rest and across the submaximal force levels. Tendon stress was greater in the instantaneous compared with the resting CSA condition, and young males had the greatest difference in the values of tendon stress between the two conditions (20 ± 4%), followed by old males (19 ± 5%), and young females (17 ± 5%). The specific force at which the difference between the instantaneous and resting CSA stress values became statistically significant was 2.6, 6.6, and 10% MVC for old males, young females, and young males, respectively. The influence of using the instantaneous compared to resting CSA for tendon stress is sex-specific in young adults, and age-specific in the context of males. The instantaneous CSA should be used to provide a more accurate measure of in vivo tendon stress in humans.

Estimation of maximal muscle electromyographic activity from the relationship between muscle activity and voluntary activation

Maximal muscle activity recorded with surface electromyography (EMG) is an important neurophysiological measure. It is frequently used to normalize EMG activity recorded during passive or active movement. However, the true maximal muscle activity cannot be determined in people with impaired capacity to voluntarily activate their muscles. Here, we determined whether maximal muscle activity can be estimated from muscle activity produced during submaximal voluntary activation. Twenty-five able-bodied adults (18 males, mean age 29 yr, range 19-64 yr) participated in the study. Participants were seated with the knee flexed 90° and the ankle in 5° of dorsiflexion from neutral. Participants performed isometric voluntary ankle plantarflexion contractions at target torques, in random order: 1, 5, 10, 15, 25, 50, 75, 90, 95, and 100% of maximal voluntary torque. Ankle torque, muscle activity in soleus, medial and lateral gastrocnemius muscles, and voluntary muscle activation determined using twitch interpolation were recorded. There was a strong log_e-linear relationship between measures of muscle activation and muscle activity in all three muscles tested. Linear mixed models were fitted to muscle activation and log_e-transformed EMG data. Each 1% increase in muscle activation increased muscle activity by a mean of 0.027 ln(mV) [95% confidence interval (CI) 0.025 to 0.029 ln(mV)] in soleus, 0.025 ln(mV) [0.022 to 0.028 ln(mV)] in medial gastrocnemius, and 0.028 ln(mV) [0.026 to 0.030 ln(mV)] in lateral gastrocnemius. The relationship between voluntary muscle activation and muscle activity can be described with simple mathematical functions. In future, it should be possible to normalize recorded muscle activity using these types of functions.NEW & NOTEWORTHY Muscle activity is often normalized to maximal muscle activity; however, it is difficult to obtain accurate measures of maximal muscle activity in people with impaired voluntary neural drive. We determined the relationship between voluntary muscle activation and plantarflexor muscle activity across a broad range of muscle activation values in able-bodied people. The relationship between voluntary muscle activation and muscle activity can be described with simple mathematical functions capable of estimating maximal muscle activity.

A deep-learned skin sensor decoding the epicentral human motions

State monitoring of the complex system needs a large number of sensors. Especially, studies in soft electronics aim to attain complete measurement of the body, mapping various stimulations like temperature, electrophysiological signals, and mechanical strains. However, conventional approach requires many sensor networks that cover the entire curvilinear surfaces of the target area. We introduce a new measuring system, a novel electronic skin integrated with a deep neural network that captures dynamic motions from a distance without creating a sensor network. The device detects minute deformations from the unique laser-induced crack structures. A single skin sensor decodes the complex motion of five finger motions in real-time, and the rapid situation learning (RSL) ensures stable operation regardless of its position on the wrist. The sensor is also capable of extracting gait motions from pelvis. This technology is expected to provide a turning point in health-monitoring, motion tracking, and soft robotics.

Linear Logistic Regression for Estimation of Lower Limb Muscle Activations

This paper addresses a technique to estimate the muscle activity from the movement data. Statistical models, such as linear regression (LR) models and artificial neural networks (ANNs), are good candidate estimation techniques. Although an ANN has a high estimation capability, it is frequently in the clinical application that a very small amount of data leads to performance deterioration. Conversely, an LR model needs fewer data, while its generalization performance is limited. In this paper, therefore, a muscle activity estimation method is proposed that uses a linear logistic regression model to improve the generalization performance. The proposed method was compared with an LR model and an ANN in verification experiments with several different conditions. The results suggest that the proposed method has a higher generalization performance than the conventional methods.

A Longitudinal Electromyography Study of Complex Movements in Poststroke Therapy. 2: Changes in Coordinated Muscle Activation

Fine motor control is achieved through the coordinated activation of groups of muscles, or "muscle synergies." Muscle synergies change after stroke as a consequence of the motor deficit. We investigated the pattern and longitudinal changes in upper limb muscle synergies during therapy in a largely unconstrained movement in patients with a broad spectrum of poststroke residual voluntary motor capacity. Electromyography (EMG) was recorded using wireless telemetry from 6 muscles acting on the more-affected upper body in 24 stroke patients at early and late therapy during formal Wii-based Movement Therapy (WMT) sessions, and in a subset of 13 patients at 6-month follow-up. Patients were classified with low, moderate, or high motor-function. The Wii-baseball swing was analyzed using a non-negative matrix factorization (NMF) algorithm to extract muscle synergies from EMG recordings based on the temporal activation of each synergy and the contribution of each muscle to a synergy. Motor-function was clinically assessed immediately pre- and post-therapy and at 6-month follow-up using the Wolf Motor Function Test, upper limb motor Fugl-Meyer Assessment, and Motor Activity Log Quality of Movement scale. Clinical assessments and game performance demonstrated improved motor-function for all patients at post-therapy (p < 0.01), and these improvements were sustained at 6-month follow-up (p > 0.05). NMF analysis revealed fewer muscle synergies (mean ± SE) for patients with low motor-function (3.38 ± 0.2) than those with high motor-function (4.00 ± 0.3) at early therapy (p = 0.036) with an association trend between the number of synergies and the level of motor-function. By late therapy, there was no significant change between groups, although there was a pattern of increase for those with low motor-function over time. The variability accounted for demonstrated differences with motor-function level (p < 0.05) but not time. Cluster analysis of the pooled synergies highlighted the therapy-induced change in muscle activation. Muscle synergies could be identified for all patients during therapy activities. These results show less complexity and more co-activation in the muscle activation for patients with low motor-function as a higher number of muscle synergies reflects greater movement complexity and task-related phasic muscle activation. The increased number of synergies and changes within synergies by late-therapy suggests improved motor control and movement quality with more distinct phases of movement.

Inverse estimation of multiple muscle activations based on linear logistic regression

This study deals with a technology to estimate the muscle activity from the movement data using a statistical model. A linear regression (LR) model and artificial neural networks (ANN) have been known as statistical models for such use. Although ANN has a high estimation capability, it is often in the clinical application that the lack of data amount leads to performance deterioration. On the other hand, the LR model has a limitation in generalization performance. We therefore propose a muscle activity estimation method to improve the generalization performance through the use of linear logistic regression model. The proposed method was compared with the LR model and ANN in the verification experiment with 7 participants. As a result, the proposed method showed better generalization performance than the conventional methods in various tasks.