The conventional framework of motor unit (MU) control assumes that MUs in a MU pool are constrained by a fixed recruitment order and a common input. This rigid control framework has been challenged by recent findings suggesting that MU activity could be flexibly modulated, potentially mediated by descending cortical inputs. In this study, rather than evaluating flexibility from the perspective of recruitment thresholds, we investigated control flexibility by assessing whether human participants can voluntarily modulate MU firing rates beyond rigid control constraints. Specifically, we examined whether participants could voluntarily modulate the firing rates of a pair of MUs from the tibialis anterior muscle during real-time feedback. Two tasks involving target reach with different visual feedback derived from the MUs' firing rates were conducted. In both tasks, there was no evidence that participants were able to change MU firing rates in a way that would violate rigid control robustly. Our findings demonstrate limited flexibility in MU control in human tibialis anterior muscle within single-session training, even when real-time MU activity feedback was provided. The results suggest that MU flexibility is not inherently present in the human lower limb.NEW & NOTEWORTHY How flexibly can the central nervous system control individual motor units? We tested whether humans can achieve independent control of pairs of motor units from the tibialis anterior muscle in two experiments with real-time visual feedback of motor unit firing rates. The results of both experiments provided no evidence of independent control.

Abstract Background Spike trains from spinal motor neurons contain low-frequency components that modulate muscle force, and higher-frequency components (above 10 Hz) that do not. The functional role of these higher-frequency components in motor control is still debated. We investigated whether mental tasks that modulate the power of cortical oscillations produce corresponding modulations in spinal motor neuron activity above 10 Hz without affecting force output. Such coupling would indicate that some higher-frequency components are not merely arising as a byproduct of force generation nor indirectly contributing to motor control, but simply reflect cortical oscillations propagating to spinal motor neurons. If voluntary power modulations of these higher-frequency oscillations do not affect force output, they could potentially serve as control signals for neural interface applications such as movement augmentation or motor neuroprostheses. Methods We recruited 15 human participants and recorded high-density electromyography signals (HD-EMG) from the tibialis anterior muscle, as well as electroencephalography (EEG) signals. The cumulative spike train (CST) was computed from the activity of spinal motor neurons decoded from HD-EMG signals. The participants performed sustained dorsiflexion concurrent with foot motor imagery, hand motor imagery, mental arithmetic, or no specific mental task. We analysed the bandpower correlation between EEG and CST signals as well as evaluated the task discriminability of CST bandpower signals with a linear classifier. Results At the intra-muscular coherence peak, we found statistically significant power correlations between CST and EEG in two separate analyses: first, when correlating across individual trials regardless of the mental task, and second, when correlating across the four mental tasks (Kendall’s $$\tau $$ coefficient $$\tau _{trial} = 0.08 \pm 0.10$$ , $$\tau _{task} = 0.33 \pm 0.19,$$ respectively; mean ± std. dev.). To evaluate the potential of the CST as a control signal, we classified the mental tasks based on CST bandpower and obtained classification accuracies slightly but significantly above chance level ( $$30\% \pm 5\%$$ ; chance level = 25%). Conclusion These results show that mental tasks can simultaneously modulate the power of cortical and spinal oscillations. This supports the notion that cortical oscillations not contributing to ongoing force control can propagate to the spinal level. We further demonstrate that mental tasks can be classified from CST bandpower, but classification performance is limited by the low signal-to-noise ratio.

We investigated whether the same modulations in spinal motor neurons parallel power modulations of cortical oscillations induced by mental tasks. We recruited 15 participants and recorded high-density electromyography signals (HD-EMG) from the tibialis anterior muscle, as well as electroencephalography (EEG) signals. The cumulative spike train (CST) was computed from the activity of spinal motor neurons decoded from HD-EMG signals. The participants performed sustained dorsiflexion concurrent with foot motor imagery, hand motor imagery, mental arithmetic, or no specific mental task. We found significant power correlations between CST and EEG across trials irrespective of the mental task and across mental tasks at the intra-muscular coherence peak (τ\_trial = 0.08 ± 0.10, τ\_task = 0.33 ± 0.19, respectively; mean ± std. dev.). CST power in beta and low-gamma bands could provide a novel control signal for neural interface applications, as power changes in these bands are not translated into actual force changes. To evaluate the potential of CST bands as a control signal, we classified the mental tasks from CST bandpower with a linear classifier and obtained classification accuracies slightly but significantly above chance level (30% ± 5%; chance level = 25%). These results show for the first time that mental tasks can modulate the power of cortical and spinal oscillations concurrently. This supports the notion that movement-unrelated oscillations can leak down from the cortex to the spinal level. We further show that mental tasks can be classified from CST, although further research is necessary to boost the classification performance to an adequate level for neural interface applications. ### Competing Interest Statement The authors have declared no competing interest.

Spinal motor neurons receive a wide range of input frequencies. However, only frequencies below ca. 10 Hz are directly translated into motor output. Frequency components above 10 Hz are filtered out by neural pathways and muscle dynamics. These higher frequency components may have an indirect effect on motor output, or may simply represent movement-independent oscillations that leak down from supraspinal areas such as the motor cortex. If movement-independent oscillations leak down from supraspinal areas, they could provide a potential control signal in movement augmentation applications. We analysed high-density electromyography (HD-EMG) signals from the tibialis anterior muscle while human subjects performed various mental tasks. The subjects performed an isometric dorsiflexion of the right foot at a low level of force while simultaneously (1) imagining a movement of the right foot, (2) imagining a movement of both hands, (3) performing a mathematical task, or (4) performing no additional task. We classified the channel-averaged HD-EMG signals and the cumulative spike train (CST) of motor-units using a filter bank and a linear classifier. We found that in some subjects, the mental task can be classified from the channel-averaged HD-EMG signals and the CST in oscillations above 10 Hz. Furthermore, we found that these oscillation modulations are incompatible with a systematic and task-dependent change in force level. Our preliminary findings from a limited number of subjects suggest that some mental task-induced oscillations from supraspinal areas leak down to spinal motor neurons and are discriminable via EMG or CST signals at the innervated muscle.

Bibian et al. show in their recent paper (Bibián et al. 2021) that eye and head movements can affect the EEG-based classification in a reaching motor task. These movements can generate artefacts that can cause an overoptimistic estimation of the classification accuracy. They speculate that such artefacts jeopardise the interpretation of the results from several motor decoding studies including our study (Ofner et al. 2017). While we endorse their warning about artefacts in general, we do have doubts whether their work supports such a statement with respect to our study. We provide in this commentary a more nuanced contextualization of our work presented in Ofner et al. and the type of artefacts investigated in Bibian et al.

Memory-augmented neural networks (MANNs) can solve algorithmic tasks like sorting. However, they often do not generalize to lengths of input sequences not seen in the training phase. Therefore, we introduce two approaches constraining the state-space of the network controller to improve the generalization to out-of-distribution-sized input sequences: state compression and state regularization. We show that both approaches can improve the generalization capability of a particular type of MANN, the differentiable neural computer (DNC), and compare our approaches to a stateful and a stateless controller on a set of algorithmic tasks. Furthermore, we show that especially the combination of both approaches can enable a pre-trained DNC to be extended post hoc with a larger memory. Thus, our introduced approaches allow to train a DNC using shorter input sequences and thus save computational resources. Moreover, we observed that the capability for generalization is often accompanied by loop structures in the state-space, which could correspond to looping constructs in algorithms.

Movement intention detection using electroencephalography (EEG) is a challenging but essential component of brain-computer interfaces (BCIs) for people with motor disabilities.Objective.The goal of this study is to develop a new experimental paradigm to perform asynchronous online detection of movement based on low-frequency time-domain EEG features, concretely on movement-related cortical potentials. The paradigm must be easily transferable to people without any residual upper-limb movement function and the BCI must be independent of upper-limb movement onset measurements and external cues.Approach. In a study with non-disabled participants, we evaluated a novel BCI paradigm to detect self-initiated reach-and-grasp movements. Two experimental conditions were involved. In one condition, participants performed reach-and-grasp movements to a target and simultaneously shifted their gaze towards it. In a control condition, participants solely shifted their gaze towards the target (oculomotor task). The participants freely decided when to initiate the tasks. After eye artefact correction, the EEG signals were time-locked to the saccade onset and the resulting amplitude features were exploited on a hierarchical classification approach to detect movement asynchronously.Main results. With regards to BCI performance, 54.1% (14.4% SD) of the movements were correctly identified, and all participants achieved a performance above chance-level (around 12%). An average of 21.5% (14.1% SD) of the oculomotor tasks were falsely detected as upper-limb movement. In an additional rest condition, 1.7 (1.6 SD) false positives per minute were measured. Through source imaging, movement information was mapped to sensorimotor, posterior parietal and occipital areas.Significance. We present a novel approach for movement detection using EEG signals which does not rely on upper-limb movement onset measurements or on the presentation of external cues. The participants' behaviour closely matches the natural behaviour during goal-directed reach-and-grasp movements, which also constitutes an advantage with respect to current BCI protocols.

The technical world of today fundamentally relies on structural analysis in the form of design and structural mechanic simulations. A traditional and robust simulation method is the physics-based finite element method (FEM) simulation. FEM simulations in structural mechanics are known to be very accurate; however, the higher the desired resolution, the more computational effort is required. Surrogate modeling provides a robust approach to address this drawback. Nonetheless, finding the right surrogate model and its hyperparameters for a specific use case is not a straightforward process. In this paper, we discuss and compare several classes of mesh-free surrogate models based on traditional and thriving machine learning (ML) and deep learning (DL) methods. We show that relatively simple algorithms (such as k-nearest neighbor regression) can be competitive in applications with low geometrical complexity and extrapolation requirements. With respect to tasks exhibiting higher geometric complexity, our results show that recent DL methods at the forefront of literature (such as physics-informed neural networks) are complicated to train and to parameterize and thus, require further research before they can be put to practical use. In contrast, we show that already well-researched DL methods, such as the multi-layer perceptron, are superior with respect to interpolation use cases and can be easily trained with available tools. With our work, we thus present a basis for the selection and practical implementation of surrogate models.

End-of-line (EoL) testing is performed to determine product quality by ensuring reliable performance. Even though low-quality products may pass EoL testing, they have a high probability of failure over time. Analyzing product usage data can help to improve EoL testing in this regard. However, current approaches do not consider usage data for this purpose. The major challenge for manufacturers is that they do not have access to comprehensive usage data for their products because customers are unwilling to provide usage data. However, manufacturers obtain some usage data from their sales and service departments i.e., contextual data. In this paper, we introduce an alternative approach to improving EoL testing when usage data from customers are missing. We discuss whether it is possible to predict low-quality products from EoL testing data when only contextual information is available (i.e., historical service data and location data of shipped products). We find that a simple, duration-based product usage threshold is sufficient to separate products affected by the production process (low-quality products) from those affected primarily by usage and environmental factors (long-term influence). Low-quality products could only be predicted by combining EoL data and contextual data. Additionally, we identify frequent patterns of maintained components to tackle the challenge of having limited data and promote user acceptance of our predictive model. Finally, we demonstrate our approach by conducting a case study in the welding industry. Our approach can identify frequent component failures and improve product reliability in countries with varying environmental conditions and rates of product usage. We expect that our findings will improve EoL testing protocols in welding and other industries while improving defect prediction models in general. (C) 2021 Elsevier B.V. All rights reserved.

OBJECTIVE:Daily life tasks can become a significant challenge for motor impaired persons. Depending on the severity of their impairment, they require more complex solutions to retain an independent life. Brain-computer interfaces (BCIs) are targeted to provide an intuitive form of control for advanced assistive devices such as robotic arms or neuroprostheses. In our current study we aim to decode three different executed hand movements in an online BCI scenario from electroencephalographic (EEG) data.APPROACH:Immersed in a desktop-based simulation environment, 15 non-disabled participants interacted with virtual objects from daily life by an avatar's robotic arm. In a short calibration phase, participants performed executed palmar and lateral grasps and wrist supinations. Using this data, we trained a classification model on features extracted from the low frequency time domain. In the subsequent evaluation phase, participants controlled the avatar's robotic arm and interacted with the virtual objects in case of a correct classification.MAIN RESULTS:On average, participants scored online 48% of all movement trials correctly (3-condition scenario, adjusted chance level 40%, alpha = 0.05). The underlying movement-related cortical potentials (MRCPs) of the acquired calibration data show significant differences between conditions over contralateral central sensorimotor areas, which are retained in the data acquired from the online BCI use.SIGNIFICANCE:We could show the successful online decoding of two grasps and one wrist supination movement using low frequency time domain features of the human EEG. These findings can potentially contribute to the development of a more natural and intuitive BCI-based control modality for upper limb motor neuroprostheses or robotic arms for people with motor impairments.

The last decade has witnessed a rapid growth of the field of exoplanet discovery and characterisation. However, several big challenges remain, many of which could be addressed using machine learning methodology. For instance, the most prolific method for detecting exoplanets and inferring several of their characteristics, transit photometry, is very sensitive to the presence of stellar spots. The current practice in the literature is to identify the effects of spots visually and correct for them manually or discard the affected data. This paper explores a first step towards fully automating the efficient and precise derivation of transit depths from transit light curves in the presence of stellar spots. The methods and results we present were obtained in the context of the 1st Machine Learning Challenge organized for the European Space Agency's upcoming Ariel mission. We first present the problem, the simulated Ariel-like data and outline the Challenge while identifying best practices for organizing similar challenges in the future. Finally, we present the solutions obtained by the top-5 winning teams, provide their code and discuss their implications. Successful solutions either construct highly non-linear (w.r.t. the raw data) models with minimal preprocessing -deep neural networks and ensemble methods- or amount to obtaining meaningful statistics from the light curves, constructing linear models on which yields comparably good predictive performance.

We show that persons with spinal cord injury (SCI) retain decodable neural correlates of attempted arm and hand movements. We investigated hand open, palmar grasp, lateral grasp, pronation, and supination in 10 persons with cervical SCI. Discriminative movement information was provided by the time-domain of low-frequency electroencephalography (EEG) signals. Based on these signals, we obtained a maximum average classification accuracy of 45% (chance level was 20%) with respect to the five investigated classes. Pattern analysis indicates central motor areas as the origin of the discriminative signals. Furthermore, we introduce a proof-of-concept to classify movement attempts online in a closed loop, and tested it on a person with cervical SCI. We achieved here a modest classification performance of 68.4% with respect to palmar grasp vs hand open (chance level 50%).

In a case study with a person with high cervical spinal cord injury, we show a first proof-of-concept on how to detect and classify different movement attempts of the same upper limb. The lesion was complete (AIS A) at level C4 and no hand function was preserved. We detected in a self-paced online setup hand open and palmar grasp with an accuracy of 68.4 % (chance level 50%).

The aim of the MoreGrasp project is to develop a non-invasive, multimodal user interface including a brain-computer interface (BCI) for control of a grasp neuroprostheses in individuals with high spinal cord injury (SCI). The first results of the ongoing MoreGrasp clinical feasibility study involving end users with SCI are presented. This includes BCI screening sessions, in which we investigate the electroencephalography (EEG) patterns associated with single, natural movements of the upper limb. These patterns will later be used to control the neuroprosthesis. Additionally, the MoreGrasp grasp neuroprosthesis consisting of electrode arrays embedded in an individualized textile forearm sleeve is presented. The general feasibility of this electrode array in terms of corrections of misalignments during donning is shown together with the functional results in end users of the electrode forearm sleeve.

Spinal cord injury (SCI) can disrupt the communication pathways between the brain and the rest of the body, restricting the ability to perform volitional movements. Neuroprostheses or robotic arms can enable individuals with SCI to move independently, improving their quality of life. The control of restorative or assistive devices is facilitated by brain-computer interfaces (BCIs), which convert brain activity into control commands. In this paper, we summarize the recent findings of our research towards the main aim to provide reliable and intuitive control. We propose a framework that encompasses the detection of goal-directed movement intention, movement classification and decoding, error-related potentials detection and delivery of kinesthetic feedback. Finally, we discuss future directions that could be promising to translate the proposed framework to individuals with SCI.