Papers

A proof-of-concept study by Aalborg University authors tested two-dimensional cursor control in 10 able-bodied participants with no prior brain-computer interface experience. Participants used either executed or imagined tongue movements, with a state machine providing up, right, down and left options. All participants achieved control with executed movements, with a completion rate of 96.0 ± 4.9%; seven achieved control with motor imagery, with a completion rate of 41.0 ± 36.2%. Average completion times for successful targets were 30.8 ± 17.6 s and 66.5 ± 26.8 s, respectively; the latter included only the seven participants with at least one successful trial. Potential muscle artifacts during executed movements were not controlled for. This was a feasibility test in able-bodied participants, not evidence of clinical efficacy in people with severe motor impairments.

Researchers from the University of Pittsburgh and Carnegie Mellon University tested whether changes in attentional load affected intracortical BCI control in two participants with tetraplegia. Participants moved and clicked a 2D cursor while completing a working memory task. EEG measures and difficulty ratings suggested increased attentional load during dual-tasking, but BCI performance remained largely stable across the dual-task conditions. P2 showed small but significant increases in trial completion time and normalized path length under mild compared with moderate load. The authors suggested that differences in signal quality between participants may have influenced the results; this remains an interpretation rather than an established causal relationship.

University of Melbourne researchers studied one participant implanted with a Stentrode endovascular brain-computer interface who attempted to handwrite English characters. They classified the attempts offline and online using neural signal trajectories over time. Across four online sessions, each containing 20-30 trials per character and 80-180 rest trials, the best-performing 3-class model achieved a mean rest recall of 87%, character recall of 65%, and character precision of 59%. Adjusting all trials to the same duration significantly improved classification compared with raw trials. The findings support further investigation of temporal information for motor decoding, but come from one participant and a small number of classes.

A collaboration involving West China Hospital of Sichuan University combined healthy neural recordings with biophysical constraints to decode motor signals affected by pathology. In a gesture classification task using high-density electrocorticography from patients with gliomas infiltrating the motor cortex, the study reported improved classification accuracy. Correlation between motor-command representations and healthy canonical representations increased from r = 0.66 to r = 0.94. The authors interpret the model's behavior as an "inverse weighting" strategy that downweights high-power pathological noise and redirects spatial attention toward functional compensatory regions. This is a proof of concept for decoding in compromised networks, not evidence of restored motor function.

A Nanyang Technological University researcher combined EEG and EMG timing information in a gesture-recognition network, reaching 87.6% and 89.2% accuracy on datasets with 12 and 20 participants. The results exceeded the strongest baselines by 9.4 and 9.1 percentage points; adaptation required no labelled target-user data. Participants were healthy young adults.

Researchers at IIT Delhi, the All India Institute of Medical Sciences and the University of Queensland recorded EEG from ten healthy participants performing curls under five resistance conditions. Classification accuracy was 97.2% in within-subject testing, while cross-subject testing reached a maximum of 93.0%. Errors mainly occurred between adjacent levels.

Researchers proposed an upper-limb motor imagery EEG decoding framework and tested its subject-independent performance on three datasets. The method combines relationships between electrodes, handcrafted time-frequency features, and spatial and temporal feature learning. For three upper-limb classes in PhysioNet, five electrodes yielded 96.01% accuracy under leave-one-subject-out validation. On BCI IV-2b, three electrodes yielded 93.35% under the same validation protocol. The authors report competitive performance against representative graph-based methods with fewer trainable parameters and EEG electrodes. These dataset evaluations do not establish online control performance or clinical benefit.

A University of Ulsan study analyzed two independent public datasets to examine when brain rhythms begin to weaken across regions during cue-paced motor imagery and when information distinguishing imagined movements emerges. Regional changes followed an ordered sequence. Class-discriminative information rose most steeply 0.60-0.75 seconds after the cue and saturated within a further 0.3 seconds. In every participant across both datasets, this transition was closer to sensorimotor desynchronization onset than to cue presentation. However, windows centered on each region's own onset classified equivalently to reassigned windows. The authors interpret the sequence as a timing landmark for analysis rather than a directly decodable feature.

Researchers built a control framework, the Intention Matrix, that classifies brain-wave signals to infer a user's intended leg movement and fuses that with data from the exoskeleton's own sensors to generate low-level control commands. Intention reading begins after the user initiates the limb movement. The authors say the results show preliminary feasibility for supporting the EXOS-1 exoskeleton in therapy; the abstract reports no participant numbers or accuracy figures.

EEG signals vary between people and drift over time, so a brain-computer interface usually has to be recalibrated for each new user. Researchers at Hangzhou Dianzi University built TSMD-Net, which learns from several existing subjects without any data from the new user, then pools what it learned about each one into a single classifier. It reached mean accuracies of 70.21%, 77.82% and 79.53% on three public EEG datasets.

Researchers combined high-resolution surface electrophysiology recordings from the first 70 patients implanted with an anatomically conformal, thin-film neural interface to build a functional atlas of the human central nervous system. The interface was positioned at more than 100 locations, with >100,000 electrode deployments spanning motor, sensory, language and other functional regions, as well as the brainstem and spinal cord. Aggregated recordings showed spatially ordered digit representations, movement-related frequency-band changes and task-modulated neural population activity. The study also demonstrated real-time neuroprosthetic control and decoding of cortical activity associated with speech, facial movements and emotional expressions. This approach combines recordings across patients rather than simultaneously reading the entire central nervous system in one person. This study is a preprint and has not been peer reviewed.

Researchers have used a generative adversarial network, an AI model that can produce synthetic data, to generate artificial primary motor cortex signals from other functionally related cortical regions and add them to training data, addressing the shortage of task-relevant signals in any single subject's recordings. On MEG data recorded while subjects performed reaching tasks, motor decoding performance rose by about 10% (p < 0.05) after augmentation, and the gain held even without real primary motor cortex signals. The method was also applied to a motor imagery BCI competition dataset to improve classification accuracy.

Researchers implanted an intracortical brain-computer interface in a participant with complete tetraplegia that reads movement intent from motor cortex to drive muscle electrical stimulation for grasping, while hand-mounted force sensors send touch information back by stimulating the sensory cortex. A separate channel pairs spinal cord stimulation with sensory input to promote neural plasticity. The participant could self-feed and pick up fragile objects. Improvements in elbow flexion strength and wrist sensation persisted after stimulation stopped, suggesting the system may be retraining neural pathways rather than only replacing them.

Researchers have proposed BrainVLA, a framework that borrows a vision-language-action model pretrained on large amounts of robot data to decode what a paralyzed user intends to do. A key obstacle for BCIs that control external devices is the scarcity of paired neural and movement data; the usual approach of learning a task-specific map from neural activity to kinematic parameters is held back by that shortage. The team first built a paired dataset of neural activity, action signals, language instructions and rendered visual observations, then adapted the OpenVLA-OFT policy to the target action space through LoRA fine-tuning. Neural activity is aligned with language representations to extract an intention signal, which the model combines with visual input to generate executable actions. On 2 neural motor datasets with different action dimensions, the framework outperformed the evaluated baselines in cross-session decoding R² and task success rate, and was data-efficient to train. The study is a preprint and has not been peer reviewed.

When a motor imagery BCI is used on a different day, EEG signals drift, yet a standard classifier keeps issuing commands regardless of confidence. Researchers at the University of Notre Dame, reporting in the Journal of Neural Engineering, added a conformal prediction gate that withholds commands when the model is unsure. Using DeepConvNet as the base classifier, the gate cut the share of wrong commands from 37.5% to 9–16% on the OpenBMI dataset (54 subjects) and from 35.5% to 6–8% on BCI Competition IV Dataset 2a (9 subjects), at the cost of issuing fewer commands. All results come from offline public datasets.

Researchers have proposed ThinkNet, a framework that keeps a test subject's data out of every decision, from data normalization to model selection and ensembling, so that EEG decoding scores for new users are not inflated. A BCI meant for new users must read their imagined limb movements from EEG without having seen their data. On the four-class task of the public BCI Competition IV-2a dataset, using 9-fold leave-one-subject-out cross-validation, the selected compact model, with only 4.9K parameters, reached 44.35±15.41% accuracy at 0.99 ms per inference. Small models of up to 25K parameters averaged 40.10%, above medium and large models at 35.09% and 34.78%, and validation-gated ensembles outperformed single-model inference. The team also found almost no correlation between validation and test scores, leaving reliable model selection a bottleneck. The study is a preprint and has not been peer reviewed; the authors note it has been accepted at the IEEE-EMBS BHI'2026 conference.

A new review surveys the past decade of signal-processing algorithms and hardware for EEG-based motor imagery brain-computer interfaces, which let users drive external devices or retrain motor function simply by imagining a movement. In these systems, users imagine moving a hand or foot while scalp electrodes record signals from the sensorimotor cortex, and algorithms translate them into commands. The authors survey the more practical algorithms of the past ten years, focusing on how machine learning and deep learning improve performance and efficiency. They also review hardware and supporting technologies, including system-on-chip designs, application-specific chips, programmable chips, wearables, the Internet of Things and augmented and virtual reality, discuss how these pair with algorithms, and outline research directions for better real-time performance. The paper is a preprint and has not been peer reviewed.

The study tested seven detection methods to assess whether motor imagery BCIs can identify movement classes absent from training rather than force them into known categories. For many users, model uncertainty was higher for familiar classes than for unfamiliar ones, undermining methods that work well in other machine learning domains. Detection performance was correlated with performance on the trained task. Deep Ensemble and Monte Carlo-Dropout models achieved an on-task AUC of 0.9 and out-of-distribution detection ability up to 0.7. The authors conclude that rejecting unfamiliar cognitive states is feasible for models and participants with high task performance.

Researchers at Lanzhou Jiaotong University proposed a hierarchical graph learning framework to decode motor imagery EEG recorded while people watch a movement in virtual reality and imagine performing it. The method splits scalp EEG into rhythmic components, selects key features, represents each trial's brain-region connectivity as a graph and classifies it with a hierarchical graph convolutional network. On the team's self-collected VR motor imagery dataset, it outperformed common baselines including EEGNet in three-class decoding. The team also found that observing plus imagining in VR strengthened connectivity among sensorimotor regions, engaging a wider brain network. The abstract does not give specific accuracy figures or the number of participants.

Researchers recorded single neurons in the motor cortex of paralyzed people with intracortical brain-computer interface implants while they watched virtual human hands, robotic hands and hand-like dots perform grasps. The more human-like the effector, the stronger the motor cortex response, both across the neural population and in individual neurons, and the effect tracked participants' own sense of human-likeness. When users attempted hand movements themselves, however, motor cortex activity stayed largely the same regardless of what was on screen, suggesting that intuitive control of external devices may not depend much on whether the device looks human.

The study introduces a model designed to reduce the impact of differences between participants' EEG signals on motor imagery classification. It combines comprehensive deep feature representation with domain adaptation to learn information shared across participants. Tests on four benchmark datasets produced average accuracies of 84.34%, 89.59%, 73.32%, and 87.75%. The authors report better performance than the state-of-the-art deep domain adaptation models included in their comparisons. Ablation experiments also assessed the two core modules. These findings concern cross-subject classification on datasets, not demonstrated online BCI control or rehabilitation benefits.

In three participants with tetraplegia imagining object manipulation, neural activity across several brain regions represented the configuration of the whole hand during imagined grasps. The posterior parietal cortex simultaneously encoded grasp and object properties, largely through distinct neuronal subpopulations. Grasp-related activity varied with motor planning versus imagined execution and with object presence versus absence; some regions also represented object shape during planning. The authors suggest that separating grasp and object information could support more stable BCI grasp control. However, the motor imagery task does not establish restored hand function or improved robotic grasp performance. The study is a preprint that has not been peer reviewed.

Researchers developed FUSE, a two-phase learning framework that combines simultaneous EEG and MEG recordings for motor imagery classification, and evaluated it on data from 20 participants. Learned representations increased median classification accuracy from 0.734 with conventional handcrafted spectral features to 0.794. The combined approach also exceeded MEG alone at 0.680 and EEG alone at 0.765, although the gain over EEG was modest. In transfer tests, more than half of cross-subject models performed within 0.01 accuracy of their subject-specific counterparts. These findings concern signal classification and representation transfer, not clinical efficacy. This study is a preprint and has not been peer reviewed.