工程技术最新文献

In recent years, brain-computer interface (BCI) technology has made significant progress in the fields of neural engineering and human-computer interaction. Among these advances, decoding upper-limb movements from electroencephalography (EEG) signals has become a key research focus. However, most existing studies concentrate on discrete classification tasks (e.g., motor imagery recognition), while the prediction of three-dimensional continuous movement trajectories still faces several major challenges. These include the low signal-to-noise ratio of EEG signals, substantial inter-subject variability that limits generalizability, and the high degrees of freedom in 3D trajectories, which increase decoding complexity. To address these challenges and improve the accuracy of decoding continuous 3D hand movement trajectories from EEG signals, this study proposes a Multi-level Multi-scale Temporal Attention Transformer framework (M3T-Attention). The model is designed to extract temporal features across multiple time scales from EEG signals and integrate them via cross-scale attention mechanisms, enabling a nonlinear mapping from 0.5 to 12 Hz EEG signals to 3D kinematic parameters (position, velocity, and acceleration). The model was trained using EEG and wrist kinematic data from the WAY-EEG-GAL dataset. Experimental results show that the proposed method achieves Pearson correlation coefficients (PCCs) of 0.8816, 0.8841, and 0.8711 on the X, Y, and Z axes, respectively, demonstrating robust prediction performance across all subjects and outperforming existing state-of-the-art approaches. In summary, through comparative experiments, statistical significance analysis, and ablation studies, we have fully verified its ability to capture neural coding patterns. It significantly enhances the decoding performance from EEG signals to movement trajectories, offering new approaches for BCI applications in complex motor control scenarios. We have made the model's source code publicly available on GitHub repository URL: https://github.com/jjspp/M3T_Attention.

In recent years, with the rapid development of Brain-Computer Interface (BCI) and Artificial Intelligence (AI) technologies, a trend of increasing integration between the two has emerged. In some practical applications, BCI has been combined with AI to enhance the overall performance of systems, including usability, user experience, and satisfaction, especially in terms of intelligent capabilities. However, this technological integration also introduces new or exacerbates existing ethical risks, such as neural privacy breaches, cross-domain misuse, and unclear system responsibility attribution. This paper discusses the novel or more severe ethical challenges arising from the fusion of BCI and AI technologies, as well as measures and strategies to address these ethical issues, calling for the establishment of more comprehensive ethical guidelines and governance frameworks. It is hoped that this paper will contribute to a deeper understanding and reflection on the ethical risks and corresponding regulations related to the integration of BCI and AI technologies.

Parkinson's disease (PD) is a neurodegenerative disease that causes extensive impacts on cognitive and motor function, hence making correct and early diagnosis essential for efficient clinical management and enhancement of the quality of life. Analysis using EEG has now become a safe and possible method for the identification of neural abnormalities in PD. Nevertheless, current models must struggle with several constraints: high false detection rates, poor generalizability across subjects, sensitivity to EEG noise pollution, and the inability to extract deep cortical representations, which have the capability to distinguish between healthy and Parkinson patterns. To alleviate these issues, the current paper proposes a novel CortiMoS-Net (Cortical Modeling with Stacked Autoencoder and MobileNet) capable of accurately detecting Parkinson's disease from EEG signals. CortiMoS-Net architecture combines deep stacked autoencoders with low-computation MobileNet convolution blocks such that low-complexity learning of complex cortical activity patterns is supplemented with computational scalability. To achieve further enhanced model convergence and optimization of learnable parameters, the present work also proposes an enhanced hybrid optimization technique named Snow Shepherd Stride Configuration Tuning (S3C-Tune). The proposed pipeline is initiated with raw EEG signal recording, preprocessing, and peak picking for the intent of artifact removal and detection of neurologically intriguing events. Model parameters are tuned by the S3C-Tune algorithm to realize maximal training accuracy. Such a pipeline hybrid enables extensive cortical modeling as well as efficient optimization and results in correct PD vs. healthy subject classification. Experimental results confirm the effectiveness of the suggested approach with better accuracy, precision, recall, and F1-score of 0.99 and minimum error rate and minimum loss of 0.01 and 0.05, respectively. The suggested model also indicates maximum prediction correctness of 0.99 and mean efficiency measure of 0.95 as compared to a large number of state-of-the-art hybrid deep learning approaches.

Electroencephalography (EEG) can objectively reflect an individual's emotional state. However, due to significant inter-subject differences, existing methods exhibit low generalization performance in emotion recognition across different individuals. Therefore, an EEG emotion classification framework based on deep feature aggregation and multi-source domain adaptation is proposed by us. First, we design a deep feature aggregation module that introduces a novel approach for extracting EEG hemisphere asymmetry features and integrates these features with the frequency and spatiotemporal characteristics of the EEG signals. Additionally, a multi-source domain adaptation strategy is proposed, where multiple independent feature extraction sub-networks are employed to process each domain separately, extracting discriminative features and thereby alleviating the feature shift problem between domains. Then, a domain adaptation strategy is employed to align multiple source domains with the target domain, thereby reducing inter-domain distribution discrepancies and facilitating effective cross-domain knowledge transfer. Simultaneously, to enhance the learning ability of target samples near the decision boundary, pseudo-labels are dynamically generated for the unlabeled samples in the target domain. By leveraging predictions from multiple classifiers, we calculate the average confidence of each pseudo-label group and select the pseudo-label set with the highest confidence as the final label for the target sample. Finally, the mean of the outputs from multiple classifiers is used as the model's final prediction. A comprehensive set of experiments was performed using the publicly available SEED and SEED-IV datasets. The findings indicate that the method we proposed outperforms alternative methods.

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.

Recent experiments have revealed that the inter-regional connectivity of the cerebral cortex exhibits strengths spanning over several orders of magnitude and decaying with distance. We demonstrate this to be a fundamental organizing feature that fosters high complexity in both connectivity structure and network dynamics, achieving an advantageous balance between integration and differentiation of information. This is verified through analysis of a multi-scale neuronal network model with nonlinear integrate-and-fire dynamics, incorporating inter-regional connection strengths decaying exponentially with spatial separation at the macroscale as well as small-world local connectivity at the microscale. Through numerical simulation and optimization over the model parameterspace, we show that inter-regional connectivity over intermediate spatial scales naturally facilitates maximally heterogeneous connection strengths, agreeing well with experimental measurements. In addition, we formulate complementary notions of structural and dynamical complexity, which are computationally feasible to calculate for large multi-scale networks, and we show that high complexity manifests for each over a similar parameter regime. We expect this work may help explain the link between distance-dependence in brain connectivity and the richness of neuronal network dynamics in achieving robust brain computations and effective information processing.

Climate change-driven single and combined abiotic stresses pose escalating threats to sustainable, climate-smart agriculture and global food security. Melatonin (MLT, a powerful plant biostimulant) has established noteworthy potential in improving stress tolerance by regulating diverse physiological, biochemical, and molecular responses. Therefore, this review delivers a comprehensive synopsis of MLT-enabled omics responses across genomics, transcriptomics, proteomics, metabolomics, miRNAomics, epigenomics, phenomics, ionomics, and microbiomics levels that collectively regulate plant adaptation to multiple abiotic stresses. We also highlight the crosstalk between these omics layers and the power of integrated multi-omics (panomics) approaches to harness the complex regulatory networks underlying MLT-enabled stress tolerance. Lastly, we argue for translating these omics insights into actionable strategies through advanced genetic engineering and synthetic biology platforms to develop MLT-enabled, stress-smart crop plants.

A crucial method for determining a person's mental health and assessing their degree of depression is depression detection. To identify depression through speech or conversation, a number of sophisticated methods and questionnaires have been created. The constraints of the current system are as follows: reduced effectiveness as a result of poor feature selection and extraction, problems with interpretability, and the difficulty of identifying depression in different languages. As a result, the proposed model is presented to offer improved accuracy and efficient performance. While adaptive threshold-based pre-processing (AdaT) is used to eliminate quiet and unnecessary information, the twinned Savitzky-Golay filter (TSaG) is used to minimize noise in the dataset. To turn the signal into an image, a Synchro-Squeezed Adaptive Wavelet Transform Algorithm (SSawT) is employed. The Singular Empirical Decomposition and Sparse Autoencoder (SiFE) model is used to extract linear and deep features. Input's deep, linear, and statistical properties are combined using the Weighted Soft Attention-based Fusion (WSAttF) model. From the fused features, the Chaotic Mud Ring Optimization algorithm (ChMR) chooses the best features. A Dilated Convolutional Neural Network (CNN) based Bidirectional-Long Short Term Memory-Bi-LSTM (DiCBiL) is used to detect different stages of depression, which lowers error rates and increases detection accuracy. The proposed method achieves 93.22% of F1-score, 93.11% precision, 93.12% recall, and 93.31% accuracy on the DAIC-WOZ original test set. During the testing time, two more datasets, namely AVEC 2019 and MELD, are used to validate the proposed performance, attaining an accuracy of 93.91% and 85.34% respectively.

Supplementary information: The online version contains supplementary material available at 10.1007/s11571-026-10411-9.

Co-transmission, the release of multiple neurotransmitters from a single neuron, is an increasingly recognized phenomenon in the nervous system. A particularly interesting combination of neurotransmitters exhibiting co-transmission is glutamate and GABA, which, when co-released from neurons, demonstrate complex biphasic activity patterns that vary depending on the time or amplitude differences from the excitatory (AMPA) or inhibitory (GABA_A) signals. Naively, the outcome signal produced by these differences can be functionally interpreted as simple mechanisms that only add or remove spikes by excitation or inhibition. However, the complex interaction of multiple time-scales and amplitudes may deliver a more complex temporal coding, which is experimentally difficult to access and interpret. In this work, we employ an extensive computational approach to distinguish these postsynaptic co-transmission patterns and how they interact with dendritic filtering and ionic currents. We specifically focus on modeling the summation patterns and their flexible dynamics that arise from the many combinations of temporal and amplitude co-transmission differences. Our results indicate a number of summation patterns that excite, inhibit, and act transiently, which have been previously attributed to the interplay between the intrinsic active and passive electrical properties of the postsynaptic dendritic membrane. Our computational framework provides an insight into the complex interplay that arises between co-transmission and dendritic filtering, allowing for a mechanistic understanding underlying the integration and processing of co-transmitted signals in neural circuits.

Supplementary information: The online version contains supplementary material available at 10.1007/s11571-025-10383-2.

As healthcare text data becomes increasingly complex, it is vital for sentiment analysis to capture local patterns and global contextual dependencies. In this paper, we propose a hybrid Swin Transformer-BiLSTM-Spatial MLP (Swin-MLP) model that leverages hierarchical attention, shifted-window mechanisms, and spatial MLP layers to extract features from domain-specific healthcare text better. The framework is tested on domain-specific datasets for Drug Review and Medical Text, and performance is assessed against baseline models (BERT, LSTM, and GRU). Our findings show that the Swin-MLP model performs significantly better overall, achieving superior metrics (accuracy, precision, recall, F1-score, and AUC) and improving mean accuracy by 1-2% over BERT. Statistical tests to assess significance (McNemar's test and paired t-test) indicate that improvements are statistically significant (p < 0.05), suggesting the efficacy of the architectural innovations. The results' implications indicate that the model is robust, efficiently converges to classification, and is potentially helpful for a wide range of domain-specific sentiment analyses in healthcare. We will examine future research directions into exploring lightweight attention mechanisms, cross-domain multimodal sentiment analysis, federated learning to protect privacy, and hardware implications for rapid training and inference.