Non-invasive BCI

Researchers have found that whether people can learn to control a device with their thoughts may depend on whether the control scheme follows the brain's own activity patterns. Participants lying in an MRI scanner controlled a video game avatar in real time by modulating activity in brain regions responsible for spatial navigation. When the researchers then changed the mapping between brain activity and avatar movement, participants could readjust and regain control if the new mapping fell along directions in which their brain activity already varied widely, known as the intrinsic manifold; when the mapping departed from that structure, they never learned it. The researchers say the principle could be used to design BCIs that are easier to learn.

A study published in Computers in biology and medicine applies cross-frequency bispectral analysis to nonlinear EEG activity during the planning and execution of natural reach-to-grasp movements. The researchers extracted magnitude- and phase-based features from complex bicoherence matrices and assessed them through classification, permutation-based feature selection, and within-subject statistics. Execution showed stronger nonlinear coupling than planning, led mainly by beta- and gamma-driven interactions. Decoding precision versus power grasps performed similarly across the two stages, suggesting that grasp-type representations emerge during planning and persist into execution. Compared with conventional analytical baselines, bispectral features provided consistent advantages for grasp-type discrimination and multiclass classification. The findings offer a new set of motor-decoding features for future brain-computer interface and neuroprosthetic research.

Women with overactive bladder often rely on drugs and behavioral training, with limited relief. A newly registered trial at The First Affiliated Hospital of Soochow University takes a different route: patients perform a task while functional near-infrared spectroscopy reads prefrontal brain activity, and an algorithm decides when to deliver transcranial magnetic stimulation, closing the loop. The interventional, parallel-design trial plans to enroll 66 women aged 18 to 75, 33 in the experimental arm and 33 in a control arm that receives open-loop sham stimulation, with the overactive bladder symptom score as the primary endpoint. No device model or results are disclosed in the registration, and recruitment has not yet started.

The General Office of China's Tianjin Municipal People's Government issued the Action Plan for Promoting Innovative Development of Brain-Computer Interfaces in Tianjin (2026-2030), document number Jin Zheng Ban Fa [2026] No. 8, dated May 26, 2026 and released on May 28. All product targets sit on the non-invasive track, while invasive systems are covered only by milestones for entering registration trials and winning market approval. By 2027 Tianjin wants two to three national-level innovation platforms, more than five non-invasive products in clinical use, more than ten new products across elderly care, education, entertainment, smart living and advanced manufacturing, 50 innovative firms, one to two national standards, and three to five industry funds. By 2030 it targets more than ten non-invasive products in large-scale clinical use, market approval for invasive products, and a fund cluster exceeding 10 billion yuan.

Researchers proposed ResDSNet, a residual depthwise-separable network with a matching preprocessing module that filters noise while keeping motor imagery features, to break what they call the "depth barrier": deepening lightweight decoders fails when residual structures are poorly designed or mismatched with preprocessing. On 3 public datasets covering 127 subjects it reached 79.36%, 84.95% and 64.13% accuracy, 3.16%, 1.59% and 8.40% higher than the best existing methods. It was tested only offline on public data, not in real-time BCI systems or patients.

Can a non-invasive brain-computer interface read motor intention and, combined with spinal cord stimulation and exoskeleton training, improve function after spinal cord injury? Xuanwu Hospital, Capital Medical University has registered a single-arm exploratory trial that plans to enroll 10 patients aged 18 to 65 with spinal cord injury. The intervention combines all three into one rehabilitation regimen. The primary outcome is the overall clinical efficacy rate 6 months after surgery; secondary outcomes include improvement in the Visual Analogue Scale for pain and the incidence of adverse events related to the trial devices.

The trial tests whether a steady-state visual evoked potential (SSVEP) brain-computer interface can drive functional electrical stimulation of swallowing muscles more effectively than conventional neuromuscular electrical stimulation. It plans to enroll 90 patients, 45 per arm, all within 6 months of a first ischemic or hemorrhagic stroke and aged 18 to 80. The primary outcome is the Standard Swallowing Assessment, with secondary measures including the Functional Oral Intake Scale and the Penetration-Aspiration Scale. First enrollment is listed for June 1, 2026.

Researchers have built a hybrid batch-stream framework for EEG-based emotion recognition, which until now has had to choose between accurate but slow offline batch processing and fast but less accurate real-time streaming. The system estimates whether the streaming module is currently reliable from its past performance, then decides probabilistically whether to use the batch or the streaming output. It achieved classification accuracies of 85%, 94% and 74% on the public DEAP, AMIGOS and SEED benchmarks, and was also tested under concept drift, when data distributions shift over time. Results come from offline evaluation on public datasets and have not been validated with real users.

ChatBCI-Assist integrates a locally deployed LLM with adaptive stopping, achieving record P300 speller online ITR of 105.2 bits/min and 30.7 CPM in semantic spelling.

Shandong Provincial Hospital in eastern China has registered a pilot trial that will test whether patients with acute ischemic stroke can begin brain-signal-driven arm training within days of onset. The planned 60 participants will practice upper-limb movements with a scalp EEG brain-computer interface that reads their intention to move and gives training feedback, compared against a sham-training group. Both groups will train for 10 days on top of standard early rehabilitation, twice a day for about 30 minutes per session. The primary outcome is the change in upper-limb motor function score 30 days after randomization. The trial is not yet recruiting and is expected to start in October 2026.

Guangdong Provincial Hospital of Traditional Chinese Medicine has registered a parallel-design interventional trial, ChiCTR2600124411, that plans to enroll 81 patients with unilateral upper-limb impairment after stroke in three groups of 27. Two groups receive intermittent theta-burst stimulation (iTBS) over the dorsolateral prefrontal cortex or the primary motor cortex, and one receives sham stimulation; all three undergo motor imagery BCI training. Primary outcomes are BCI decoding accuracy, the Fugl-Meyer Assessment and brain network topology. The trial was registered on May 12, 2026 and is not yet recruiting.

Researchers in Chongqing, a municipality in southwestern China, have proposed SPD-DANN, a method that aims to decode motor imagery, the brain signals produced when a person imagines moving a hand or foot, in new users without per-person calibration. Because EEG varies between people and drifts over time, BCIs typically need fresh data from each new user. SPD-DANN represents EEG features on a symmetric positive definite (SPD) matrix manifold, a mathematical space that better fits their nonlinear structure, and uses adversarial learning to extract features shared across individuals. Tested on 4 BCI datasets, the method outperformed several existing unsupervised domain adaptation techniques, the researchers said. The abstract gives no accuracy figures, and all tests were offline.

A randomized, sham-controlled trial will pair a non-invasive brain-computer interface with MRI-guided multichannel temporal interference stimulation in patients whose migraines resist medication. It plans to enroll 78 people aged 18 to 65, split evenly between real and sham stimulation, with headache frequency as the primary endpoint. The registration lists only design and bibliographic fields, so device details and actual enrollment remain unknown and efficacy has yet to be read out.

The Chinese Clinical Trial Registry (ChiCTR) has logged a multicenter randomized controlled trial testing whether brain-computer interface-based mindfulness meditation relieves the symptoms of chronic tinnitus. Registered on April 23, 2026 as ChiCTR2600123258, the trial has not begun enrolling patients, and its design, sample size and sponsor must be read from the original registration record. It is the first multicenter registration to combine BCI with mindfulness meditation for chronic tinnitus, moving the approach from single-center exploration toward multicenter validation.

The trial combines a brain-computer interface with robotic mirror therapy into a single intervention, while the control arm receives conventional comprehensive rehabilitation only. It enrolls patients aged 30 to 80 with unilateral upper-limb motor impairment 1 to 6 months after ischemic or hemorrhagic stroke, whose lesions are confined to the basal ganglia. The primary endpoint is the Fugl-Meyer Assessment of the Upper Extremity, with EEG, functional MRI and the Modified Ashworth Scale among secondary measures. The planned enrollment is 40 patients, with first enrollment set for May 1, 2026.

Researchers at Xidian University built DisCo-Former, a Transformer framework for single-trial rapid serial visual presentation (RSVP) EEG decoding with three components guided by neurophysiological priors, then found its attention consistently collapsed: attention maps went nearly uniform and value-projection weights shrank toward 0. Stripping out the Transformer encoder left DisCo-MLP, a pure multilayer perceptron that matched or beat the Transformer version across two datasets and three evaluation regimes, with within-subject mean AUCs of about 0.94 to 0.98. For RSVP-EEG, the authors argue, modeling the signal's structure matters more than architectural complexity. The study was published in the International Journal of Neural Systems on April 10, 2026.

People's Daily Online reported that BrainCo completed an RMB 2 billion Series B+ round in January 2026, the largest single BCI financing in China at the time. The report did not disclose the lead investor, a complete investor list or the post-money valuation. The round indicates a capital-intensive expansion phase across BrainCo's non-invasive and medical product lines.

The FDA cleared Zeto's New Wave System on March 13, 2026, finding it substantially equivalent under submission number K260455 through the special 510(k) pathway. The agency classifies it as a Full-Montage Standard Electroencephalograph, a Class II device under regulation 882.1400, with neurology as the assigned review panel. That category covers standard electroencephalograph hardware acquiring a complete montage of scalp channels.

Researchers built STAND-Net, a spiking neural network that cleans blink, muscle and other artifacts from EEG and computes only when the signal changes, combining a leaky integrate-and-fire encoder-decoder, a dilated-convolution residual backbone and an attention mechanism that locates artifacts from neuron firing patterns. Against state-of-the-art methods it raised signal-to-distortion ratio by more than 3.7 dB across artifact types and used 97.98% less power than comparable deep neural networks; BCI classification on the cleaned signals was 6.64% more accurate. The results come from the paper's abstract, with no data from wearers.

Researchers designed an assistive Boccia ramp controlled by a brain-computer interface so that children with severe motor impairments who cannot speak can take part in the Paralympic sport. Six caregivers took part in semi-structured interviews and six children completed a newly developed 21-item survey; the children showed keen interest in playing Boccia with a BCI. The study also maps current barriers and facilitators to sport participation, and the issues that must be resolved before such technology can be rolled out.

Researchers in Germany, the Netherlands and Japan argue in an opinion piece that brain-computer interface design should be rebuilt around ideomotor theory, which treats voluntary action as driven by internally represented sensory outcomes. BCI research has made remarkable technical progress but remains limited in scope, the authors write, typically relying on motor and visual cortex signals in a narrow range of patient populations, and they describe this underused framework as a principled basis for next-generation interfaces that align more closely with the brain's own intentional and action-planning architecture. Reorganizing BCIs around the purpose of an action, meaning the user's goals and anticipated effects, would be a more intuitive, generalizable and scalable path, they suggest, and advances in neural recording and artificial intelligence-based decoding of sensory representations make the shift feasible and timely, potentially easing persistent usability and generalizability problems in BCI design.

Researchers designed a paradigm in which the same visual stimulus elicits five cognitive processes (reaching, intercepting, reaching in sequence, reaching after attending, and withholding a reach) and distinguished them from scalp EEG with 91.72% ± 6.10% accuracy, using a regularized linear discriminant analysis classifier with event-related spectral perturbation features. In the sequence reach task, temporal features separated normal from catch trials with 77.96% ± 7.03% accuracy. The analysis was offline; the team plans to improve decoding and extend it to online use.

Researchers have proposed MSHANet, a deep learning model that reads imagined limb movements from scalp EEG, a signal often used in rehabilitation after stroke or spinal cord injury. The paper notes that motor imagery EEG is nonlinear and varies widely between people, which limits decoding accuracy. The model combines spatiotemporal feature extraction, interactive-head self-attention, dynamic squeeze-and-excitation attention and a temporal convolutional network. On three public datasets it reached within-subject accuracies of 83.56%, 89.75% and 75.66%, and cross-subject accuracies of 69.93%, 81.85% and 79.67%. A separate encoding method that writes electrode positions into the input raised within- and cross-subject accuracy on BCI-2a by 2.83% and 2.91%. All results come from offline public data and have not been validated in real time on patients.

A study published in January 2026 proposes RUNet, a motor imagery EEG decoding framework designed to skip the calibration session BCIs usually need before each use, since EEG varies across sessions and between people because of individual and environmental factors. RUNet is first pre-trained without labels on resting-state EEG to learn session- and subject-invariant features, then retrained on task data, using Riemannian manifold mappings to suppress cross-session and cross-subject drift. It reached an average cross-session accuracy of 87.19% and a cross-subject accuracy of 68.09% on BCI Competition IV 2a, and a cross-subject accuracy of 78.14% on PhysioNet. All results come from offline dataset tests and have not been validated in real-time use.

BrainCo's FocusXin Medical Edition combines a single-channel prefrontal EEG device registered as 浙械注准20232071898 with ADHD rehabilitation software registered as 浙械注准20252211934. The Class II system is intended for supplementary rehabilitation training in children aged 6 to 12 with ADHD. Its defined medical use should not be generalized into approval for an unspecified consumer BCI product.

The FDA cleared Hippoclinic's HippoMind, version 1.0, on December 3, 2025, finding it substantially equivalent under submission number K251881 through the traditional 510(k) pathway. The agency classifies it as Non-Normalizing Quantitative Electroencephalograph Software, a Class II device under regulation 882.1400, with neurology as the assigned review panel. Unlike normalizing tools, non-normalizing quantitative EEG software reports an individual's own EEG metrics without reference to a population database.

The FDA cleared Naox Technologies' NX01 on November 25, 2025, finding it substantially equivalent under submission number K251550 through the traditional 510(k) pathway. The agency classifies it as a Reduced-Montage Standard Electroencephalograph, a Class II device under regulation 882.1400, with neurology as the assigned review panel. Reduced-montage systems record with fewer electrodes than a conventional clinical montage, trading spatial resolution for faster setup and the possibility of wearable form factors.

Nihon Kohden's EEG-1260A Neurofax system, along with its amplifier unit and photic stimulator, received FDA 510(k) clearance on October 9, 2025. The device is a Class II medical device in the neurology specialty, with software for quantitative EEG analysis. The traditional 510(k) clearance indicates substantial equivalence to a legally marketed device.

The FDA cleared Forest Devices' SignalNED System (Model RE) on September 3, 2025, finding it substantially equivalent under submission number K251726 through the traditional 510(k) pathway. The agency classifies it as a Reduced-Montage Standard Electroencephalograph, a Class II device under regulation 882.1400, with neurology as the assigned review panel. Forest Devices has applied reduced-montage EEG to prehospital stroke identification, acquiring interpretable signals in settings such as ambulances.

Neuracle Medical Technology (Shanghai) Co., Ltd. ran a clinical trial in China evaluating the safety and efficacy of its implantable wireless brain-computer interface, NEO, in patients with tetraplegia caused by spinal cord injury. The trial is registered on ClinicalTrials.gov as NCT06990412 and is listed as completed; its stated primary purpose was to provide the basis for product registration. NEO is designed to let patients control external devices with brain signals.