NeuroXess announced a RMB 700 million Series B round co-led by Guotou Juli, Shenzhen Capital Group, and Qinzhi Capital. The company says proceeds will support multicenter registration trials and regulatory approval work for its fully implantable, fully wireless, fully functional BCI system, manufacturing facilities, next-generation products, and BCI–AI integration. It reports that the registration trial has started at Huashan Hospital, Fudan University, and that its planned factory covers 14,300 square meters, targeting production capacity on the scale of 10,000 systems. These clinical and manufacturing updates are company disclosures: starting a registration trial does not mean device approval, and planned capacity is not current output.
Why it matters
The planned spending covers both registration trials and manufacturing, distinct steps toward medical-market entry. Clinical progress, regulatory decisions, and production milestones need to be assessed separately; closing the funding round does not establish that those goals have been achieved.
A review examines clinical safety evidence for intracortical, cortical-surface and cerebral-venous brain-computer interfaces and proposes an assessment framework spanning preimplant evaluation through end-of-use. The authors identify barriers to comparison: small and sometimes overlapping cohorts, inconsistent exposure denominators, incomplete long-term surveillance and limited retrieval data. Their framework combines a common dataset with architecture-specific assessments and an event-triggered process for confirmation, attribution, escalation and outcome recording. They stress that no single test establishes chronic safety. Prospective multicentre testing, independent adjudication and registry follow-up are needed before the approach can inform formal guidance.
Why it matters
Safety assessment extends beyond surgical complications and device changes to the loss of functions participants may depend on. The review includes this participant-level outcome, calling for device status and functional changes to be tracked together over time.
Researchers proposed an integrated dual framework for motor imagery and steady-state visually evoked potential BCIs, focusing on rehabilitation and robotics research. Developed through a scientific literature review, it organizes 20 components into experimental and BCI system sections, covering both non-technical phases critical to study validity and technical architecture. A supporting taxonomy and checklist help structure terminology and check whether study elements are present. Documentation assessment in three BCI studies demonstrated the framework's applicability as a design and reporting tool, rather than rehabilitation efficacy or robot-control performance.
Why it matters
The framework places ethical considerations and user characterization alongside system components. It offers a concrete lens for comparing BCI studies: whether non-technical phases are documented as fully as the technical architecture.
A systematic review and network meta-analysis synthesized 16 randomized controlled trials comparing BCI-triggered functional electrical stimulation, BCI-controlled robotic rehabilitation and conventional rehabilitation for motor recovery after stroke. Electrical stimulation showed a benefit over conventional rehabilitation, while the effect of robotic rehabilitation remained uncertain. However, no trial directly compared the two BCI approaches, and their indirect comparison did not establish a difference. The findings therefore do not establish that electrical stimulation is superior to robotics; confidence in that comparison is very low, and direct head-to-head trials are needed.
Why it matters
Different findings against conventional rehabilitation do not establish a difference between the two BCI approaches. The star-shaped evidence network makes this distinction explicit and explains why treatment rankings remain exploratory.
Researchers from Kunming University of Science and Technology and collaborating institutions proposed a theoretical system for basic BCI research, formalizing BCIs as closed-loop probabilistic information-processing models through information theory and statistical decision-making. The work addresses system decodability, performance boundaries and dynamic constraints. It organizes scientific hypotheses, neuroscience principles, fundamental theorems and basic laws, derives four results that can guide experimental design, and uses motor imagery BCIs to illustrate application boundaries. The authors aim to provide a unified analytical benchmark and support a shift from experience-driven algorithm optimization toward theory-driven system analysis and design.
Why it matters
Rather than reporting a more accurate decoder, the work builds a theoretical structure for examining when systems are decodable and what constrains performance. Motor imagery illustrates application boundaries; it should not be read as evidence of validated performance or clinical benefit.
Researchers from Kocaeli University and collaborating institutions proposed an adaptive multi-domain feature-fusion framework to distinguish left-hand motor imagery, right-hand motor imagery, mental arithmetic and rest from functional near-infrared spectroscopy signals. Using subject-independent leave-one-subject-out cross-validation on an open-access dataset of 29 participants, the framework achieved 95.6% mean accuracy, compared with 90.7% for temporal features alone, 95.0% for cepstral features alone and 95.2% for direct concatenation. Accuracy gains over direct concatenation ranged from 0.2 to 0.7 percentage points for every participant. Most of the gain over temporal features came from the cepstral domain, with adaptive weighting adding a small, consistent improvement.
Why it matters
The study separates the contribution of adding another feature domain from that of adaptive weighting. Cepstral features account for most of the gain, rather than the weighting mechanism delivering a large improvement on its own.
A descriptive feasibility study involving Huashan Hospital, Fudan University recruited 90 participants: 30 healthy young adults, 31 healthy older adults and 29 people with spinal cord injury. It recorded 32-channel EEG during continence motor attempts. Groups showed broadly consistent sensorimotor rhythm changes. During suppression, an older-versus-younger high-frequency difference was significant, while the spinal cord injury group’s similar trend was not. Frontal high-frequency findings may reflect residual artifacts. This characterizes intent-related EEG rather than testing restoration of bowel or bladder function.
Why it matters
Discernible cortical signals during motor attempts do not mean that urination or defecation function has recovered. The study describes EEG patterns across groups to inform future intent-decoding and pelvic nerve modulation research, rather than reporting treatment effects.
Researchers at Shaanxi University of Science & Technology developed DWDANet to address EEG differences between people that limit cross-subject generalization. The framework combines multi-scale spatiotemporal feature modeling with dynamically weighted domain alignment. Cross-subject accuracy reached 70.61%, 81.43% and 64.38% on BCI IV-2a, BCI IV-2b and SHU-MI, respectively, outperforming the existing methods compared on all three datasets. The study also reported within-subject results. These evaluation conditions differ: higher within-subject accuracy should not be treated as performance for a new user, and the dataset findings do not establish rehabilitation efficacy.
Why it matters
Beyond extracting spatiotemporal EEG features, the framework dynamically weights multiple source domains during alignment. Its focus is on using data from different sources to improve cross-subject classification, rather than only recognition within one participant's data.
Researchers at Kunming University of Science and Technology reviewed the mechanisms and outcome evaluation of neurofeedback for post-traumatic stress disorder. They compared EEG, real-time functional magnetic resonance imaging and functional near-infrared spectroscopy in feedback targets, spatiotemporal resolution and clinical applicability. Neurofeedback provides real-time information about brain activity to guide patients in regulating abnormal neural patterns. Using clinical scales, neurophysiological measures and functional recovery indicators, the review discusses its clinical position relative to first-line treatments and potential synergy with psychological interventions. The authors identify inadequate control designs and non-standardized training parameters as ongoing limitations; therapeutic potential should not be treated as established efficacy.
Why it matters
The review brings symptom scales, brain-activity measures and functional recovery into one evaluation framework, distinguishing neural changes from improvements in daily functioning. Proposed synergy with psychological interventions is not established additional efficacy.
A review surveys tools for studying the collective activity of neural populations and discusses how low-dimensional structure can help explain sensory coding, working memory, decisions, and sensorimotor transformations. The authors emphasize that cognitive signals can occupy separable or near-orthogonal population dimensions, reducing interference without requiring distinct cell classes: individual neurons in a shared population can still respond to multiple kinds of information. The review also considers recurrent neural networks as testbeds for mechanistic hypotheses and intracortical brain-computer interfaces as an application in which population geometry becomes a control signal. This is a review, not a new clinical efficacy study.
Why it matters
A useful distinction concerns interpretation: a subspace defined by an analysis method is not necessarily a neural manifold with reproducible geometry or topology. The review separates these concepts when discussing population activity and BCI control signals.
