Simulating Human Upper and Lower Limb Balance Recovery Responses Using Nonlinear Model Predictive Control.

The ability to generate predictive dynamic simulations of human movement using optimal control has been a growing point of interest in the design of medical/assistive devices, e.g. robotic exoskeletons. Despite this, many disseminated simulations of whole-body tasks, such as balance recovery, neglect the role of the upper body instead focusing on postural joints, e.g. ankle, knees, hips. Thus, the purpose of the current study was to use a novel nonlinear model predictive control (NMPC) approach to assess how actuated upper limbs, as well as different individual performance (optimality) criteria, can shape simulated reactive balance recovery responses. A sagittal biomechanical model of a young adult standing was designed and actuated via nonlinear muscle torque generators (rotational single-muscle equivalents). Forward dynamic simulations of balance recovery (NMPCdriven) following an unexpected support-surface perturbation were generated for each unique combination of selected performance criteria (6 total), perturbation direction (forward and backward), and arm joints free/locked. The observed joint trajectories provide insight into the emergence of human elements of postural control from individual optimality criteria, e.g. hip-ankle strategies emerge from single-joint regulation. Quantitative analysis of performance improvements with the arms free suggest that whether arm responses emerge in the simulations may be dependent on the problem's initial guess. Future work should focus on testing further performance criteria and improving NMPC as a model of the nervous system.

The torque-frequency relationship is impaired similarly following two bouts of eccentric exercise: No evidence of a protective repeated bout effect.

High-intensity eccentric exercise can lead to muscle damage and weakness. The 'repeated bout effect' (RBE) can attenuate these impairments when performing a subsequent bout. The influence of eccentric exercise-induced muscle damage on low-frequency force production is well-characterized; however, it is unclear how eccentric exercise and the RBE affect torque production across a range of stimulation frequencies (i.e., the torque-frequency relationship). We investigated the influence of an initial (Bout 1) and repeated bout (Bout 2) of eccentric exercise on the elbow flexor torque-frequency relationship. Eleven males completed two bouts of high-intensity eccentric elbow flexions, 4 weeks apart. Torque-frequency relationships were constructed at baseline and 0.5, 24, 48, 72, 96, and 168 h following both bouts via percutaneous stimulation at 1, 6, 10, 20, 30, 40, 50, and 100 Hz. Serum creatine kinase activity, self-reported muscle soreness, and isometric maximum voluntary contraction torque indirectly inferred the presence of muscle damage following Bout 1, and attenuation of muscle damage following Bout 2. Torque amplitude at all stimulation frequencies was impaired 30 min following eccentric exercise, however, torque at lower (1-10 Hz) and higher frequencies (40-100 Hz) recovered within 24 h while torque across the middle frequency range (20-30 Hz) recovered by 48 h. No between-bout differences were detected in absolute or normalized torque at any stimulation frequency, indicating no protective RBE on the elbow flexor torque-frequency relationship.

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.

Power loss is attenuated following a second bout of high-intensity eccentric contractions due to the repeated bout effect's protection of rate of torque and velocity development.

High-intensity unaccustomed eccentric contractions result in weakness and power loss because of fatigue and muscle damage. Through the repeated bout effect (RBE), adaptations occur, then damage and weakness are attenuated following a subsequent bout. However, it is unclear whether the RBE protects peak power output. We investigated the influence of the RBE on power production and estimated fatigue- and damage-induced neuromuscular impairments following repeated high-intensity eccentric contractions. Twelve healthy adult males performed 5 sets of 30 maximal eccentric elbow flexions and repeated an identical bout 4 weeks later. Recovery was tracked over 7 days following both bouts. Reduced maximum voluntary isometric contraction torque, and increased serum creatine kinase and self-reported soreness indirectly inferred muscle damage. Peak isotonic power, time-dependent measures - rate of velocity development (RVD) and rate of torque development (RTD) - and several electrophysiological indices of neuromuscular function were assessed. The RBE protected peak power, with a protective index of 66% 24 h after the second eccentric exercise bout. The protection of power also related to preserved RVD (R2 = 0.61, P < 0.01) and RTD (R2 = 0.39, P < 0.01). Furthermore, the RBE's protection against muscle damage permitted the estimation of fatigue-associated neuromuscular performance decrements following eccentric exercise. Novelty: The repeated bout effect protects peak isotonic power. Protection of peak power relates to preserved rates of torque and velocity development, but more so rate of velocity development. The repeated bout effect has little influence on indices of neuromuscular fatigue.

Modelling the dynamic margins of stability for use in evaluations of balance following a support-surface perturbation.

The dynamic margin of stability provides a method that captures the center of mass (CoM) state (position-velocity) in relation to the base of support (BoS). However, the model upon which this concept was derived does not consider how the inertial characteristics of forced support-surface perturbations would influence balance control. Within the current article, the inverted pendulum model was restructured to account for fixed, piecewise accelerations of the BoS. From this logic, two variations of the adjusted margin of stability, each maintaining a similar definition of extrapolated CoM, are proposed; one ignoring horizontal ground contact and inertial forces applied to the BoS, the other incorporating these forces. Unique within the proposed models is the time-variant BoS boundaries that depend on the perturbation applied. Verification of the solution for each model is provided, along with a comparison of obtained values to previous methods of defining CoM position-velocity stability metrics using a computational model and optimal control. For the simpler model variation (ignoring forces), we also assessed how CoM position and perturbation parameter selection over/underestimate the predicted maximal permissible velocity. The results of these analyses suggest that factors which increase the acceleration impulse decrease the difference between the two models; the opposite was observed for factors increasing displacements between the CoM and BoS boundary. Lastly, use of the proposed adjusted margin of stability within an experimental data set highlights the ability of our model to predict instability (stepping strategies; negative margin of stability) relative to the use of the extrapolated CoM alone.

Repeated Exposure to Forward Support-Surface Perturbation During Overground Walking Alters Upper-Body Kinematics and Step Parameters.

Locomotion requires both proactive and reactive control strategies to maintain balance. The current study aimed to: (i) ascertain upper body postural responses following first exposure to a forward (slip) support-surface perturbation; (ii) investigate effects of repeated perturbation exposure; (iii) establish relationships between arms and other response components (trunk; center of mass control). Young adults (N = 11) completed 14 walking trials on a robotic platform; six elicited a slip response. Kinematic analyses were focused on extrapolated center of mass position (xCoM), bilateral upper- and forearm elevation velocity, trunk angular velocity, and step parameters. Results demonstrated that postural responses evoked in the first slip exposure were the largest in magnitude (e.g., reduced backward stability, altered reactive stepping, etc.) and preceded by anticipatory anterior adjustments of xCoM. In relation to the perturbed leg, the large contra- and ipsilateral arm responses observed (in first exposure) were characteristically asymmetric and scaled to the degree of peak trunk extension. With repeated exposure, xCoM anticipatory adjustments were altered and in turn, reduced posterior xCoM motion occurred following a slip (changes plateaued at second exposure). The few components of the slip response that persisted across multiple exposures did so at a lesser magnitude (e.g., step length and arms).

Modeling margin of stability with feet in place following a postural perturbation: Effect of altered anthropometric models for estimated extrapolated centre of mass.

BACKGROUND: Maintaining the centre of mass (CoM) of the body within the base of support is a critical component of upright balance; the ability to accurately quantify balance recovery mechanisms is critical for many research teams. RESEARCH QUESTION: The purpose of this study was to investigate how exclusion of specific body segments in an anthropometric CoM model influenced a dynamic measure of postural stability, the margin of stability (MoS), following a support-surface perturbation. METHODS: Healthy young adults (n = 10) were instrumented with kinematic markers and a safety harness. Sixteen support-surface translations, scaled to ensure responses did not involve a change in base of support, were then issued (backwards, forwards, left, or right). Whole-body CoM was estimated using four variations of a 13-segment anthropometric model: i) the full-model (WFM), and three simplified models, ii) excluding upper limbs (NAr); iii) excluding upper and lower limbs (HTP); iv) pelvis CoM (CoMp). The CoM calculated for each variant was then used to estimate extrapolated CoM (xCoM) position and the resulting MoS within the plane of postural disturbance. RESULTS: Comparisons of simplified models to the full model revealed significant differences (p < 0.05) in MoS for all models in each perturbation condition; however, the largest differences were following sagittal plane based perturbations. Poor estimates of WFM MoS were most evident for HTP and CoMp models; these were associated with the greatest values of RMS/maximum error, poorest correlations, etc. The simplified models provided low-error approximates for frontal plane perturbations. SIGNIFICANCE: Findings suggest that simplified calculations of CoM can be used by researchers without reducing MoS measurement accuracy; however, the degree of simplification should be context-dependent. For example, CoMp models may be appropriate for questions pertaining to frontal plane MoS; sagittal plane MoS necessitates inclusion of lower limb and HTP segments to prevent underestimation of postural stability.

Do perturbation-evoked responses result in higher reaction time costs depending on the direction and magnitude of perturbation?

To date, little work has focused on whether cognitive-task interference during postural response execution is influenced by the direction and/or magnitude of the perturbation applied. Hypothetically, the increased difficulty associated with a backward loss of balance could necessitate increased allocation of cognitive resources to counteract destabilizing forces. The current study investigated these relationships using a paradigm in which individuals performed a cognitive task (auditory Stroop task during quiet stance; baseline condition). In certain trials, a translation of the support surface was concurrently evoked (magnitude: small or large; direction: forward or backward) which required a postural response to maintain balance. Ten healthy young adults completed four blocks of these experimental trials (26 randomized trials/block). Postural stability during balance recovery was evaluated using the margin of stability (MoS), while Stroop task performance was based on reaction time cost (RTC) and differences between experimental conditions. Results showed no effect of perturbation direction on RTC, but there was an observed MoS increase at peak extrapolated center of mass excursion following a small perturbation evoked concurrently with the cognitive task. No effect of cognitive-task performance was detected for MoS during stepping strategies (followed large perturbations). Instead, increased RTC were observed relative to the fixed base of support responses. In general, young adults adopted a "posture-first" strategy, regardless of perturbation direction, reinforcing the importance of cognition in the maintenance of upright balance.

Slip and Trip Perturbations During an Object Transport Task Requiring a Lateral Change in Support.

The ability to counteract destabilizing external forces while simultaneously executing a complex task presents a novel way to ascertain one's ability to generate adaptive postural control responses to avoid a potential fall. In this study, participants performed an upper limb object transport task requiring a lateral change in support on a robotic platform that could remain fixed in space or translated (mimicking a slip or trip perturbation). No significant stability differences were observed at initial recovery step between slip and trip perturbations. Variability measures were greatest during the trip perturbations; though stability was at its greatest level preceding these perturbations. These results will aid in the design of future studies that will investigate adaptive postural control responses generated by older adults when executing similar, ongoing complex upper body tasks interrupted by a destabilizing support surface perturbation.