工程技术最新文献

Traffic accidents usually result from driver's inattention, sleepiness, and distraction, posing a substantial danger to worldwide road safety. Advances in computer vision and artificial intelligence (AI) have provided new prospects for designing real-time driver monitoring systems to reduce these dangers. In this paper, we assessed four known deep learning models, MobileNetV2, DenseNet201, NASNetMobile, and VGG19, and offer a unique Hybrid CNN-Transformer architecture reinforced with Efficient Channel Attention (ECA) for multi-class driver activity categorization. The framework defines seven important driving behaviors: Closed Eye, Open Eye, Dangerous Driving, Distracted Driving, Drinking, Yawning, and Safe Driving. Among the baseline models, DenseNet201 (99.40%) and MobileNetV2 (99.31%) achieved the highest validation accuracies. In contrast, the proposed Hybrid CNN-Transformer with ECA attained a near-perfect validation accuracy of 99.72% and further demonstrated flawless generalization with 100% accuracy on the independent test set. Confusion matrix studies further indicate a few misclassifications, verifying the model's high generalization capacity. By merging CNN-based local feature extraction, attention-driven feature refinement, and Transformer-based global context modeling, the system provides both robustness and efficiency. These findings show the practicality of using the suggested technology in real-time intelligent transportation applications, presenting a viable avenue toward reducing traffic accidents and boosting overall road safety.

Little is known about whether direct and vicarious rewards affect bilingual language control in social learning. We used a dual-electroencephalogram (EEG) to simultaneously record the effects of direct and vicarious rewards on language control when bilinguals switched between their two languages. We found that both direct and vicarious rewards elicited more switch behavior. On an electrophysiological level, although both direct and vicarious rewards elicited Reward-positivity and Feedback-P3 when receiving reward outcomes, direct rewards induced greater reward effects than vicarious rewards. In addition to an N2 effect in language switching, vicarious rewards elicited more pronounced LPCs relative to direct rewards. More important, in the alpha band, there was a predictive effect of behaviors on rewards in binding vicarious rewards and language switching activities. These findings demonstrate that both direct and vicarious rewards influence language control during language selection.

Spiking neural networks (SNNs) have gained significant attention for their biological plausibility, event-driven operation, and low power consumption, establishing them as a leading model for processing event stream data. However, current models often oversimplify neuronal dynamics to balance computational cost and performance. To address this limitation and enhance the dynamical behavior of spiking neurons, this paper introduces two key innovations. First, inspired by biological autaptic connections and memristive devices, we propose the memristive autapse (M-Autapse), a self-connection mechanism that enables adaptive modulation of a neuron's membrane potential. Second, recognizing the need for attention mechanisms that match SNNs' spatio-temporal nature, we design a spatio-temporal synergistic attention (STSA) mechanism to bolster simultaneous focus on both temporal and spatial dimensions of input data. Extensive experiments on the neuromorphic speech benchmarks SHD and SSC validate our methods. On SHD, our model demonstrates performance competitive with the state-of-the-art, while also achieving strong results on the SSC dataset.

Cell membrane-derived vesicles play essential roles in intercellular communication, material transport, and waste disposal. Despite their biomedical and industrial potential, isolating extracellular vesicles from natural sources remains technically challenging, limiting purification efficiency and scalability. This study introduces cell membrane extrusion as an alternative approach to optimize the production of cell membrane-derived vesicles (CSMs), from eukaryotic and prokaryotic cells. CSMs, generated from HeLa and SH-SY5Y cells exhibited a distinctive cup-shaped morphology and sizes of 151.36 ± 72.36 nm, and 416.86 ± 108.49 nm at 20 °C by DLS respectively, showing remarkable thermal stability at 4-70 °C range. Furthermore, loaded vesicles interacted with mammalian cells and achieved successful cargo internalization. CSMs were also produced from E. coli membranes, forming unilamellar vesicles of approximately 100 nm, as observed by Cryo-TEM. These vesicles displayed an inverse correlation between vesicle size and thermal stability and efficient cargo incorporation detected in 85% ± 3% of CSMs. However, under tested conditions, no interaction with prokaryotic cells occurred, and consequently, no delivery of the loaded molecule was observed. Overall, thesefindings highlight the potential of generating cell membrane-derived nanovesicles through extrusion, offering a promising strategy to mimic extracellular vesicles for innovative biomedical and industrial applications, including targeted drug delivery system.

The nonlinear fractional soliton neuron model is used in many complicated domains, including fluid mechanics, applied science, neuroscience, nonlinear dynamics, mathematical physics, engineering, biosciences, plasma physics, and geology. It demonstrates how nonlinear waves propagate. This article explores the the fractional nonlinear soliton neuron model, which can play a crucial role in understanding complicated phenomena in neuroscience. This model describes how axons generate and transmit action potentials using a thermodynamic theory of nerve transmission. Signals flowing through the cell membrane (CM) are believed to be single sound pulses, also known as solitons. In this paper we applied the modified Sardar sub-equation method to investigate the exact traveling wave solutions to fractional nonlinear soliton neuron model. In order to discuss the wave patterns of the model, the principal nonlinear equation is transformed to an ordinary differential equation by the usual traveling-wave transformation. The given method provides precise solutions in the form of hyperbolic, trigonometric, and exponential, which results in a well-defined system of solitary wave structures. Consequently, we obtain distinct types of solutions, containing bright solitons, dark solitons, bright-peakon mixtures, peakon and anti-peakon shapes, cuspon and compacton, compacton-kink, kink and anti-kink, dark-bright, bell-shaped solitons, singular periodic, and isolated periodic. The obtained solutions are described in terms of their physical behavior with the help of several visual tools, such as 2D and 3D surface plots, density plots, and contour plots, which have been used to display how the fractional terms reshape the wave profiles, alter their strength, and influence their overall stability. The findings provide a better understanding of the nonlinear wave propagation in neuron-inspired systems and suggest that it may be relevant in any optical media, plasma environment, and computational modeling of wave-based dynamics. Besides these analytical solutions, there is also a modulation instability study, which brings out the fact that minimal perturbation can cause instability of the underlying wave forms. This additional understanding can be used to explain the sensitivity of the model and the circumstances within which wave patterns are stable or start to increase.

The present study aimed to dissect the underlying mechanism of the Gei Herba (LBZ) in the treatment of type 2 diabetes mellitus (T2DM) based on the gut microbiota and network pharmacology strategies. Thirty-one compounds from LBZ were screened and 187 corresponding targets were identified through the database and literature screening. Multiple disease-associated gene and target databases were used to screen and obtain 726 T2DM-related targets. The gutMGene v1.0 database was used for searching metabolites and targets of gut microbiota. 168 overlapping targets of LBZ, T2DM and the gut microbiota were matched and used to build a protein-protein interaction network and perform enrichment analysis. By bioinformatics analysis of microbiota-pathway-target-compound network, the PI3K-Akt signalling pathway was identified, and potential gut probiotics, such as Eubacterium limosum and Lactobacillus paracasei, were found. Subsequently, the core active compounds in LBZ bind to the core targets were verified by molecular docking, all of which exhibited good binding affinity. Finally, four compounds, eleutheroside A, gallic acid, kaempferol, and protocatechuic acid, were found to conform to Lipinski's rule and be non-toxic, showing great potential in the treatment of T2DM. These findings suggest the potential mechanism of LBZ in the treatment of T2DM and the core compounds, such as eleutheroside A may be the promising candidates for T2DM treatment.

Biological sensory systems achieve remarkable robustness through efficient cross-modal integration, yet replicating this in artificial spiking neural networks (SNNs) remains a challenge. Inspired by Mayer's multi-channel learning cognitive theory, we present CrossModal-Associated-SNN, a neuro-inspired framework that synergistically integrates visual and auditory information via Spike-Timing-Dependent Plasticity (STDP) clustering and associative learning. The architecture employs a multi-channel, multi-network design for modality-specific processing, followed by a cross-channel complementary strategy that refines decision-making through associative signals. Evaluated on small-sample benchmarks (MNIST3K and Spoken-MNIST3K), the model demonstrates superior generalization and robustness in multi-modal classification. Compared to the single-channel baseline (91% accuracy), the dual-channel dual-network improved visual and auditory recognition to 93% and 84%, respectively, with a fusion accuracy of 94%. The dual-channel triple-network architecture further maximized performance, attaining 96% (visual), 90% (auditory), and a peak 97% cross-modal fusion accuracy. These results suggest that cooperative shallow micro-networks, akin to biological small-neuron ensembles in superficial brain regions, offer a potentially energy-efficient alternative for multimodal tasks processing. CrossModal-Associated-SNN represents a critical step toward mimicking human sensory cognitive integration, offering a biologically plausible solution for energy-efficient, multi-modal intelligent systems.

An external electric field can induce polarization effects in the excitable media and the inner diffusion of intracellular ions is modified, accompanying by appropriate shape deformation in the media. Polarization effect supports orderliness when the medium suffers from external applied electric field, and appropriate shape deformation is induced in the flexible media, including cardiac tissue and some chemical reactions. In this article, we extend an excitable medium model by incorporating both memristive electromagnetic induction and shape deformation, and further establish an equivalent network framework to investigate the resulting collective dynamics. In one-layer excitable media, the media size (L _x ·L _y ) maintains a constant, and shape deformation seldom changes the media size, as a result, the two diffusion coefficients (D _x , D _y ) become complementary with time. The local kinetics of the memristive media is discussed, and the reaction-diffusion equations are converted into equivalent coupled neural network for exploring collective behaviors, including wave propagation and synchronization stability. The shape deformation induces changes of the diffusion coefficients in opposite way, with one direction enhanced and the other weakened, and the wave stability is modified because asymmetrical diffusion in the media. Incorporation of the memristive current and shape deformation into the excitable can better reflect the physical effect in the excitable media well.

In neuronal dysfunction, there is a disruption of synthesis and release of neurotransmitters including dopamine. Oxidative stress is a major contributor to neuronal dysfunction. Oxidative stress affects tetrahydrobiopterin (BH₄), an essential cofactor in the synthesis of several neurotransmitters such as dopamine and nitric oxide. BH₄ supports the activity of enzyme tyrosine hydroxylase (TH) and initiates dopamine synthesis. Studies have reported contradictory results on neuronal dysfunction and disease progression following treatments to reduce oxidative stress and BH₄ supplementation. In this study, we developed a computational model of TH biochemical pathway in dopaminergic nerve cells. Using this model, we quantitatively analyzed the impact of reduced BH₄ synthesis and oxidative stress on L-dopa and dopamine synthesis. The base case concentration of L-dopa was 382.5 nM and cytosolic dopamine was 12.4 nM. The results showed a significant decrease in the concentrations of L-dopa and dopamine with reduction in BH₄ synthesis. The presence of oxidative stress further exacerbated these decreases in concentration. The mechanistic analysis suggests that this computational model provides an initial framework for evaluating how BH₄-related perturbations influence species concentrations in the TH biochemical pathway in dopaminergic nerve cells. Our findings indicate that under the simulated conditions, BH₄ supplementation produced short-term changes in species concentrations and that sustained improvement in nerve cell dysfunction likely requires a multi-target approach that simultaneously enhances de novo BH₄ synthesis and reduces cellular oxidative stress.

Supplementary information: The online version contains supplementary material available at 10.1007/s11571-026-10486-4.