工程技术最新文献

Closed-loop neuromodulation aims to adjust therapeutic stimulation in real time based on ongoing neural or physiological signals. Despite growing clinical adoption, most implementations rely on heuristic rules rather than a principled systems-and-control formulation. This paper, motivated by discussions from the Brain Theory Seminar (Shanghai, March 2025), develops such a formulation around seven fundamental questions-mechanism (Q1), plant nature (Q2), state measurement (Q3), actuation (Q4), modeling (Q5), objectives (Q6), and constraints (Q7)-and, for each, provides a knowledge-based review synthesizing current understanding together with a prospective scientific opinion on unresolved issues. Five recurring themes unify the seven questions: (i) nonstationarity as the default operating condition, (ii) structural partial observability and under-actuation, (iii) closed-loop confounding between stimulation and measurement, (iv) the primacy of hard constraints over unconstrained optimization, and (v) the necessity of layered governance separating performance seeking from safety enforcement. We argue that the neural plant is fundamentally different from classical engineered systems in ways that reshape what can be sensed, modeled, actuated, and verified; accordingly, we reframe therapeutic goals from setpoint tracking toward set-based regulation within a therapeutic window, and we treat safety, ethics, and accountability not as external add-ons but as architectural primitives that define the admissible design space. We close with a discussion synthesizing system-level barriers and near-term architectural directions, including bidirectional brain-computer interfaces, hybrid learning-and-control pipelines with independent safety supervision, and digital twins as regulated test harnesses.

This study investigates how electromagnetic induction and astrocytic modulation jointly influence firing dynamics and synchronization in excitatory-inhibitory neuronal networks. Using a computational model that integrates pyramidal neurons, interneurons, and astrocytes with memristor-based electromagnetic feedback and calcium-dependent signaling, we demonstrate that electromagnetic induction generally suppresses neuronal firing, but this effect can be bidirectionally modulated by astrocytic feedback depending on intracellular calcium levels. Key findings reveal the non-linear dependence of neuronal responses on astrocytic calcium thresholds and [Formula: see text] dynamics; the distinct roles of excitatory, inhibitory, and astrocytic coupling in regulating network synchrony and chimera states; and the existence of parameter regimes where astrocytic feedback is overridden under strong electromagnetic induction. These results highlight the critical interplay between electromagnetic and glial mechanisms in shaping network activity, offering insights for models of neural synchronization and potential therapeutic strategies for epilepsy and related disorders.

Electroencephalography (EEG) enables the direct measurement of stroke-related brain activity, which is crucial for objectively evaluating therapeutic effects of neuromodulatory interventions in stroke. Numerous studies indicate that EEG spectra comprise a mixture of periodic oscillations and aperiodic background; however, how aperiodic activity, distinct from oscillatory components, characterizes and quantifies external stimulation remains unclear. In this paper, we parameterized the power spectra of EEG recorded from 30 stroke subjects with acupuncture stimulation and investigated the changes of oscillatory and aperiodic brain activity following acupuncture stimulation. For the oscillatory components, we found that acupuncture increased both the power and center frequency of alpha oscillations in the left fronto-parietal cortex, indicating a shift toward faster and more strongly synchronized activity in the lesioned regions. Next, analysis of aperiodic brain activity demonstrated widespread decreases in the aperiodic exponent and offset, most prominently in the ipsilesional hemisphere. These globally and spatially coherent changes suggest that aperiodic dynamics are highly sensitive to acupuncture, reflecting coordinated changes in the background spectral organization. Furthermore, analyses of both periodic and aperiodic features revealed that acupuncture improved interhemispheric symmetry, with particularly pronounced effects observed in alpha center frequency, alpha power, and the aperiodic exponent. Collectively, this study extracted oscillatory and aperiodic spectral features to characterize EEG activity, identified alpha center frequency, alpha power, and the aperiodic exponent as potential biomarkers of acupuncture-induced changes in stroke, and provided evidence for the modulatory effects of peripheral acupuncture stimulation on brain activity.

The polar environment harbors extremely rich and valuable natural product resources. In this study, a strain of Antarctic fungus, Geomyces sp. wnf-18c, in Antarctic soil was isolated in a laboratory. This fungus produces a purplish-red pigment whose hue closely resembles that of the high-end pigment carminic acid. However, approximately 50 % of the pigment remains within the fungal hyphae, restricting subsequent separation and extraction. In this study, the deletion of the key cell wall protein-encoding genes CHS3 and AGS1 enhanced pigment secretion. It also considerably inhibited the strain's growth, making it difficult to substantially increase the total pigment yield. To overcome these hurdles, the xylose-inducible system from Trichoderma reesei was heterologously expressed in Antarctic fungi, and the CRISPR-Cas9 system was placed under the control of this induction system. The exogenous addition of xylose as a signal to initiate genetic editing achieved the spatiotemporal regulation of the target gene-activating gene editing that restricts growth only after the completion of the strain's growth phase. The spatiotemporal regulation of key cell wall proteins significantly mitigated growth inhibition, greatly enhanced pigment secretion, and raised the total pigment yield. This discovery advances our understanding of the spatiotemporal specificity and dynamic regulation of microorganisms, offering novel strategies for exploiting polar microbial resources and reducing industrial production costs.

The rapid development of motor imagery brain-computer interface (MI-BCI) technology has introduced a novel mode of human-machine interaction, drawing significant interest. However, due to the nonlinearity and nonstationarity of motor imagery signals, this communication method faces challenges such as low recognition accuracy and poor stability, which severely hinder the practical application of MI-BCIs. Riemannian geometry techniques, leveraging their non-Euclidean properties, have shown promise in addressing motor imagery signal pattern recognition. Traditional Riemannian geometry methods, however, still suffer from limitations, such as requiring manual feature extraction and resulting in suboptimal classification performance. This paper proposes Aug-SPD-AttentionNet, a deep Riemannian geometry learning framework for MI-BCI classification that combines lagged augmented covariance representations, a learnable high-dimensional embedding layer, and self-attention operations defined on the SPD manifold. The proposed framework employs multiple deep Riemannian learning blocks to extract manifold-aware features and uses a Riemannian self-attention mechanism to capture long-range dependencies among different SPD representations. Aug-SPD-AttentionNet was evaluated on BCI Competition III Dataset IVa, BCI Competition IV Dataset I, and BCI Competition IV Dataset 2a, achieving mean classification accuracies of [Formula: see text], [Formula: see text], and [Formula: see text], respectively. Ablation experiments further demonstrate that the proposed components collectively improve classification performance, increasing the mean accuracy across the three evaluated subjects from 70.60 to 82.62%, corresponding to a 12.03-percentage-point improvement over the baseline SPDNet.

Acupuncture has shown clinical benefits in post-stroke rehabilitation, yet its central neural mechanisms remain incompletely understood. In particular, how cortical activation, inter-regional functional connectivity, and brain network topology are modulated during electroacupuncture stimulation has not been systematically characterised. This study aimed to investigate the central effects of verum electroacupuncture (VE) applied to the non-dominant hand in healthy adults. In a randomized, crossover design, functional near-infrared spectroscopy (fNIRS) data were collected from 24 participants during VE and sham electroacupuncture (SE) sessions. Cortical activation was assessed using a general linear model (GLM), functional connectivity (FC) was analysed using lagged cross-correlation, and brain network topology was examined using graph-theoretical metrics. No significant differences in cortical activation were observed between VE and SE. During the needles retention phase, VE significantly reduced functional connectivity between R-MPFC and right primary somatosensory cortex (R-PSC), and between middle MPFC and R-PSC. Graph-theoretical analysis revealed decreased clustering coefficient and increased small-worldness under VE. These findings suggest that VE may optimise brain networks through an connectivity-topology mechanism. Brain networks shifted from excessive localisation toward higher integration. This study establishes a baseline central response profile in healthy individuals and provides theoretical support for the application of electroacupuncture in post-stroke upper-limb rehabilitation. Trial registration: ITMCTR, TRN: ITMCTR2025000595, Registration date: 19 March 2025.

Event-related potentials (ERPs) play an important role for building EEG-based brain computer interfaces (BCIs). However, due to the complex and varied spatio-temporal characteristics of ERPs across subjects, data distribution across subjects a very important issue to be solved for constructing calibration-free BCIs. To achieve calibration-free ERPs recognition, we propose a Domain Generalized Feature Embedded Learning (DGFEL) method. First, we align the ERPs of each existed subject based on covariance centroids. Then, we enhanced the aligned samples based on xDAWN filter and extract spatio-temporal features. Finally, the spatio-temporal features are further generalized by the decomposed adversarial loss, and we construct a neural network embedding backbone to implement features generalization across subjects. The proposed method has been systematically validated on two benchmark EEG-based ERP datasets, and its classification performance surpasses several state-of-the-art methods as well as deep learning models. Moreover, it effectively captures robust features from existed source subjects, and can be generalized to new subjects without accessing target ERP samples. Our method therefore provides a novel selection to construct calibration-free ERP-BCIs.

Accurate and reproducible electroencephalography (EEG)-based classification of dementia remains a key challenge in computational neurodiagnostics. The open-access AHEPA dataset has become the most commonly used benchmark for Alzheimer's disease (AD) and Frontotemporal dementia (FTD) classification, yet reported results vary widely due to methodological inconsistencies. This study presents the first systematic and quantitative benchmark review of all published machine learning approaches applied to the AHEPA dataset. Forty-six studies were reviewed and stratified into three validity tiers, with Validity 1 representing the highest methodological rigor and Validity 3 the lowest.According to their evaluation rigor: (1) subject-level validation (e.g., Leave-One-Subject-Out cross-validation, LOSO-CV), (2) subject-level train/test splits, and (3) epoch-level k-fold cross-validation. Performance metrics were normalized across classification problems. The analysis revealed that methodological rigor is inversely correlated with reported accuracy: for AD versus Cognitively Normal controls, mean accuracy decreased from 90.81% overall to 82.11% in Validity-1 studies; for FTD versus controls, accuracy dropped from 86.53% to 75.18%. Linear regression analyses demonstrated that weaker validation protocols were associated with systematic increases of 7-10% points in reported accuracy, explaining more than half of the observed performance variance. Deep and hybrid models reported the highest nominal accuracies, but under proper validation, traditional algorithms performed comparably, indicating that data leakage often drives apparent improvements. The review also highlights the lack of cross-configuration generalization and the urgent need for adaptive, montage-independent methodologies. Overall, this benchmark establishes the first reproducible reference framework for EEG-based dementia classification on the AHEPA dataset, providing quantitative baselines and validity criteria against which all future studies should be evaluated.

Autism spectrum disorder (ASD) has resisted reproducible case-control characterization in resting-state functional connectivity, leaving open whether reported group differences reflect genuine biology or methodological variability. We introduce a geometric framework that represents functional connectomes on the symmetric positive-definite Riemannian manifold and quantifies individual deviation as tangent-space dispersion, complemented by normative modeling of graph topology. Using the multi-site Autism Brain Imaging Data Exchange (ABIDE I) cohort (N=1035 subjects), we evaluate this descriptor under stringent robustness conditions and report three findings. First, individual connectome geometry is a stable, individually unique descriptor. The magnitude of a subject's deviation is highly reliable across split halves (intraclass correlation ICC[Formula: see text]) but is dominated by an acquisition-site scale, whereas its high-dimensional shape identifies subjects within-session with 98.9% accuracy once that scale is removed. Second, the case-control difference in dispersion is null and, by two one-sided equivalence tests against Cohen's d = 0.2, statistically equivalent to zero in all ABIDE I analysis cohorts (full cohort d = -0.007, N = 1035; strict cohort d = +0.038, N = 635); a nominal single-atlas effect (dispersion ratio 1.05, p = 0.046) does not survive multi-atlas reanalysis. Third, acquisition site explains 69% of dispersion variance while diagnosis explains 0.0001%; deviations form no discrete subtypes and do not predict symptom severity (minimum p=0.15 across tested graph-deviation predictors). Geometric dispersion is therefore a reproducible, individually unique connectome descriptor but not a case-control biomarker of ASD, which helps to explain the inconsistency of connectomic findings in autism.

Biometric traits are unique characteristics of an individual's body or behavior that can be used for identification and authentication. Biometric authentication uses unique physiological and behavioral traits for secure identity verification. Traditional unimodal biometric authentication systems often suffer from spoofing attacks, sensor noise, forgery, and environmental dependencies. To overcome these limitations, our work presents multimodal biometric authentication integrated with the characteristics of electroencephalograph (EEG) signals and handwritten signatures to enhance security, efficiency, and robustness. EEG-based authentication uses the brainwave patterns' intrinsic and unforgeable nature, while signature recognition demonstrates an additional behavioral trait for effectiveness. Our system processes EEG data of an individual with 14-channel readings, and the signature with the images ensures a seamless fusion of both modalities.Combining physiological and behavioral biometrics, our approach will significantly decrease the risk of unimodal authentication, including forgery, spoofing, and sensor failures. Our system, evaluated on a dataset of 30 subjects with genuine and forged data, demonstrates a 97% accuracy. Designed for small organizations, the modular structure, low computation algorithms, and simplicity of the hardware promote deployment scalability.