Papers

Endovascular brain-computer interfaces avoid opening the skull, but the device must be threaded through cerebral veins that vary from person to person. Researchers trained an autonomous robotic controller to navigate from the right internal jugular vein to the superior sagittal sinus. It succeeded in 14 of 20 physical runs in fluoroscopy-guided vessel phantoms, while performance dropped on an unseen anatomy in simulation. A failure-prediction module flagged most simulated failures in advance. The work is a preprint and has not been peer reviewed.

Researchers found that a decoder trained on peripheral physiological signals (heart rate, heart rate variability, respiration, electrodermal activity and pupil diameter) read arousal more accurately than EEG, reaching a within-subject area under the curve (AUC) of 93.0%, versus 85.2% for an EEG-only decoder and 79.8% for the original filter-bank common spatial pattern decoder. The offline analysis used a public simulated-flight dataset. Only the peripheral decoder reproduced the inverted-U relationship between arousal and performance, and EEG neurofeedback did not change overall arousal trajectories, suggesting its effect may come from regulation at task-critical moments. The study is a preprint and has not been peer reviewed.

When a user looks at targets flickering at different frequencies, the visual cortex produces EEG responses at matching frequencies, known as steady-state visual evoked potentials (SSVEP), which a brain-computer interface can use to infer the intended target. Such systems usually need calibration data, and more data means more user fatigue. Researchers propose SCR-FTSDNN, a deep convolutional network that moves EEG into the frequency domain, then extracts temporal and spatial features with added spatial and channel reconstruction modules. Tested on the public Benchmark and BETA datasets, the authors report higher recognition accuracy and stability than several existing methods, especially with limited calibration data. The abstract gives no specific figures, and the results come from offline tests.

Would drivers accept a car that reads their brain signals and steps in when they are overloaded? In an online experiment, 1201 participants each viewed 1 of 12 animated driving scenarios. In each scenario, a wearable, non-invasive brain-computer interface detected high workload and triggered assistance. Acceptance varied systematically across scenarios. It was higher when the system supported driving-related functions and when the driver shown had a disability. The 2 automation activation modes tested made no difference.

A study combines EEG data with multilingual text to classify emotions, since EEG alone is prone to noise while language models read sentiment in words but miss implicit changes in the brain. The author reports 89.7% accuracy for the fused model, versus 75% for EEG alone and 79% for text alone, with an F1-score of 0.88 and an average inference latency of 180 ms, enough for real-time recognition. The abstract does not state the number of participants or the datasets used.

Researchers have proposed CortexBridge, a lightweight adapter that lets EEG foundation models, usually pretrained on a single fixed electrode layout, handle any montage. It combines EEG features with electrode and brain-atlas coordinates to map signals into a shared cortical latent space before passing them to the pretrained model, which is left unchanged. On 5 BCI datasets from the public MOABB benchmark, tested with 3 frozen foundation models, EEGPT, LaBraM and CBraMod, it improved performance in 13 of 15 evaluations, raising balanced accuracy by an average of 0.80%, 0.70% and 3.26% respectively, with a maximum gain of 13.02% on 12-class steady-state visually evoked potential classification. The 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 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.

Color vision tests usually rely on what the subject reports, while a brain-computer interface can read the brain's response to color differences directly: the more two flickering light sources differ in perceived color, the stronger the steady-state visual evoked potential in the EEG. Existing systems sample color space evenly and repeat measurements at each point, which is accurate but slow. MetaIDNet, a closed-loop system, instead uses Gaussian process regression and a neural network to decide where to measure next. Offline, it made assessment 70% faster at the same accuracy; online, it improved accuracy by 40% at the same speed. The offline and online experiments involved only 10 and 8 participants.

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 NEUROTOKEN, a single network that uses conditional flow matching to score candidate voices and identify, from EEG, which speaker a listener is attending to in a noisy room, the so-called cocktail-party problem. On the public KU Leuven, DTU and NJU datasets with 5-second windows, it raised per-segment accuracy by 9% to 16% over the strongest non-generative baseline and narrowed across-subject variance by about 3x. The authors also report that the 95% to 97% accuracies often cited for decoding attention direction fall by 17% to 45% once training and test trials are strictly separated. The work is a preprint and has not been peer reviewed.

A meta-analysis of 26 randomized controlled trials covering 1100 stroke patients found that brain-computer interface (BCI) rehabilitation improved upper-extremity motor function compared with control interventions, but showed no significant benefit for lower-limb function or walking tests. A dose-response analysis suggested an inverted U-shaped relationship, with the peak effect at approximately 48 min per session.

A review of brain-computer interface (BCI) research in children finds preliminary evidence that non-invasive systems are feasible for those above 5 years of age, but the supporting studies are small and short. It compares five paradigms (P300, motor imagery, SSVEP, fNIRS and hybrid systems) and four applications: communication aids, motor rehabilitation, ADHD attention training and autism social-emotional intervention. Evidence is firmer for communication and motor rehabilitation in cerebral palsy than for ADHD and autism.

Speech brain-computer interfaces aim to restore communication by turning neural activity tied to speech, language or communicative intent into text, synthesized voice or avatar movements. This review argues they are not simple neural-to-text decoders but clinical systems that change over time, in which neural representations, recording hardware, decoding models, language models, feedback and the user's own learning interact. The authors say the next generation should be judged not only on offline accuracy but on stability across sessions, calibration burden, latency, the ability to signal uncertainty, usability and safeguards against decoding words a user did not mean to say. The paper is a preprint and has not been peer reviewed.

Researchers have released BrainNet Studio, a single toolkit that takes brain network data from construction and feature extraction through predictive modeling and visualization. Brain networks map structural and functional links between brain regions, but most existing tools handle only static networks and miss how those links change over time. The toolkit bundles 27 algorithms spanning deep learning, graph neural networks and spatiotemporal sequence models, and uses a large language model to generate connectivity summaries that researchers can check. The developers say it can support cognitive neuroscience, exploratory brain disorder studies and brain-computer interface research. The work is a preprint and has not been peer reviewed.

Can watching movements help stroke survivors retrain their arms? This scoping review screened 386 records and included 22 studies of upper-limb action observation therapy that measured EEG. Mu and beta rhythms over sensorimotor areas consistently desynchronized while patients watched movements, a sign that motor-related brain regions were engaged, and multi-week training produced sustained increases that often coincided with better upper-limb function scores. The authors see promise in BCI-based closed-loop therapy that uses EEG to adjust training in real time, but note that widely varying protocols make results hard to compare across studies.

An offline EEG study of 15 participants found that on-screen flickering targets stepping up or down in frequency can be recognized as brain-control commands, but they did not classify more accurately than conventional fixed-frequency flicker. Mean normalized accuracies were 89.7% for fixed frequency, 85.6% for ascending and 89.5% for descending patterns, a statistically insignificant difference. Low-frequency stimulation at 6–10 Hz reached 90.8%, versus 80.6% at 41–45 Hz. Sequences of three consecutive commands averaged 91.5 ± 9.6%, with accuracy falling as sequences grew longer. The researchers see the main value in more flexible design of consecutive commands and say larger samples and online closed-loop testing are needed.

Researchers have developed an adaptive version of Artifact Subspace Reconstruction (ASR) that removed 53.1% of blink artifacts from EEG recordings, compared with 24.0% for the original ASR and 26.7% for a limited-memory variant. EEG signals drift over time and are often contaminated by blinks and other artifacts, a problem BCIs must solve for long-term use in real-world settings. Standard ASR suppresses artifacts automatically based on variance but relies on a static covariance model derived from calibration data, which becomes less robust as signals drift. The new method tracks signal structure with an exponential moving average, combining short-term covariance estimates with slow recursive updates so the model adapts on a fast and a slow timescale at once. It was tested on 24-channel EEG from 10 subjects across different cognitive states. The trade-off was larger reconstruction and spectral deviation. The study is a preprint and has not been peer reviewed.

Researchers testing 3 EEG foundation models, general-purpose models pretrained on large amounts of EEG data, on 235 subjects excluded from training found that a lightweight adapter trained for each individual raised mean balanced accuracy by 1.5 to 5.4 percentage points over the shared population model. Every new brain-computer interface user needs calibration, and a personally tuned model may be more accurate either because it captures the individual's traits or simply because the underlying general model is stronger. Personal adapters also beat adapters borrowed from other people by 2.3 to 7.3 points. When the population model's training was scaled up fourfold, the personalization gain stayed positive but narrowed. Low-cost calibration using only a few labeled samples gave inconsistent results. The study is a preprint and has not been peer reviewed.

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.

Researchers have built AutoBCI, a framework in which AI agents design EEG decoding models, and report that it slightly outperformed the strongest existing model across 14 datasets. Building a decoder that works well across many EEG tasks usually takes repeated trial and error. AutoBCI splits the job between 2 agents: one generates and refines architectures, while the other forecasts final performance from early learning curves to pick out candidates worth training further. Across 14 EEG datasets covering motor imagery, emotion recognition and sleep staging, the architecture found with Claude Opus 5.5 reached an average test balanced accuracy of 64.16%, slightly above the 63.87% of REVE, the strongest baseline. Looking at only 10 training epochs, the forecasting agent cut prediction error from 2.20 to 1.36 percentage points. 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 at Bishop’s University compared visual decoding using EEG, MEG, intracranial EEG, and fMRI. Individual differences were reproducible within an instrument, but external paired cohorts did not show a common participant ranking across instruments. Classifiers were tested on recording runs and stimulus identities excluded from training. Attention, behavior, and peripheral signals affected recording quality but did not explain most stable differences. Intracranial recordings also linked sharper local category selectivity to easier decoding. The authors interpret these differences as reflecting the relationship between the individual, the distinction being represented, and the instrument, rather than a fixed personal trait alone.

A collaboration involving Peking University converted dynamic natural movie frames into spatiotemporal patterns of optogenetic stimulation in awake mice, then reconstructed movie frames from responses in the primary visual cortex. Comparing natural viewing with artificial stimulation, they found that stimulation evoked broadly comparable cortical dynamics and aligned representations. A deep learning model reconstructed frames from cortical signals under both conditions. The animal study offers a proof of concept for evaluating how visual cortical prostheses might translate images into stimulation patterns. It does not establish that stimulation produced the corresponding visual experience in mice or restored vision in humans.

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.