Search

This study investigated the impact of LNP023 on the AMPK/mTOR signaling pathway in lupus nephritis (LN) and its effects on autophagy and oxidative stress. A mouse model of LN was established, and renal injury was confirmed by assessing various LN markers, including antinuclear antibody, ds‐DNA, anti‐Sm antibody, and others. Mice were treated with LNP023, the AMPK activator AICAR, or the AMPK inhibitor dorsomorphin. Renal injury and fibrosis were evaluated using HE and Masson staining. Expression levels of AMPK, mTOR, LC3, Beclin1, and p62 were assessed by immunohistochemistry and Western blot. Oxidative stress and inflammatory markers were measured by polymerase chain reaction and enzyme‐linked immunosorbent assay. LN mice exhibited low AMPK/p‐AMPK and high mTOR/p‐mTOR levels, alongside significant renal injury, fibrosis, reduced autophagy, and elevated oxidative stress. LNP023 treatment improved these parameters, with enhanced effects when combined with AICAR. Conversely, dorsomorphin reversed LNP023's therapeutic benefits. The complement factor B inhibitor LNP023 promotes kidney health in LN mice by mediating the AMPK/mTOR pathway, promoting autophagy, and attenuating oxidative stress.

3-D air-like metamaterials (ALMs) appear omnidirectionally invisible in free space, which enables the possibility of material existence without involving any scattering to arbitrary incident electromagnetic (EM) waves. Due to their peculiar property, ALMs are quite interesting in microwave and optical engineering. However, the existing ALMs used to work with some predefined conditions, e.g., either incident polarization or operation bandwidth are limited, which prevents their implementation in wide practical applications. In this article, we present the design and measurement of a slab-type ALM, which is polarization-free and works in a couple of radar bands. This ALM is made by utilizing a full polarization compensation in 3-D, i.e., a multilayered structure with triangular constituents. The designed ALM possesses constitutive parameters identical to air in both X and Ku bands, adopting double Lorentz resonances. In the full-wave simulations, the ALM shows air-like scatteringless at around 8 and 13.5 GHz. In the experimental measurements, the ALM is nearly scatteringless in the same bands, with incident angles varying from 0° to 60° for both the vertical and horizontal polarizations. In-depth analysis shows that zero phase delay is introduced to the propagating waves, with the ALM being present. To the best of our knowledge, it is the first attempt to simultaneously break the polarization and bandwidth limitations of ALMs. The designed ALM would be a good candidate for facilitating superior antenna radomes, EM windows, as well as through-wall detections and communications.

In the absence of an official statement about a rumor, people may expose the truth behind such rumor through their responses on social media. Due to the varying relevance of responses in exposing hidden suspicious points within a rumor claim, it is crucial to prioritize those with higher relevance, rather than considering every responding tweets. As for the multi-modal rumor detection, an effective approach for evaluating relevance is aligning responses with the different modalities of the rumor claim in a fine-grained manner. However, owing to the substantial volume of response tweets, it is both costly and redundant to align all responses with the multi-modal claim. In this paper, we propose a novel two-stage model, termed Focal Reasoning Model (FoRM), to select critical responses for multi-modal rumor detection. More specifically, our FoRM consists of two primary elements: coarse-grained selection and fine-grained reasoning. The coarse-grained selection component employs post-level features of responses to initialize a relevant score for each. Based on these scores, we preserve the responses with higher scores as the candidate ones for subsequent reasoning. Within the fine-grained reasoning component, we develop a relation attention module to investigate fine-grained relationships, specifically token-to-token and token-to-object connections, between the preserved responses and the multi-modal claim, with the goal of discovering valuable clues. Extensive experiments have been conducted on three real-world datasets, and the results demonstrate that our proposed model outperforms all the baselines.

AbstractIn this article, we introduce notions which are called property (c*) and property (M3*) for semitopological groups. We show that if Gis a regular semitopological group with a q-point, property (c*) and Sm(G) ≤ ω, then Gis topologically isomorphic to a subgroup of the product of a family of first-countable M1-semitopological groups (Nagata semitopological groups). In the third part of this article, we give an internal characterization of subgroups of product of firstcountable M1-semitopological groups. A semitopological (paratopological) group Gis topologically isomorphic to a subgroup of the product of a family of first-countable M1-semitopological (paratopological) groups if and only if Gsatisfies the T0separation axiom and has property (M3*).

Freezing of gait (FOG) leads to an increased risk of falls and limited mobility in individuals with Parkinson's disease (PD). However, existing research ignores the fine-grained quantitative assessment of FOG severity. This paper provides a double-hurdle model that uses typical spatiotemporal gait features to quantify the FOG severity in patients with PD. Moreover, a novel multi-output random forest algorithm is used as one hurdle of the double-hurdle model, further enhancing the model's performance. We conduct six experiments on a public PD gait database. Results demonstrate that the designed random forest algorithm in the double-hurdle model–hyperparameter independence framework achieves outstanding performances with the highest correlation coefficient (CC) of 0.972 and the lowest root mean square error (RMSE) of 2.488. Furthermore, we study the effect of drug state on the gait patterns of PD patients with or without FOG. Results show that “OFF” state amplifies the visibility of FOG symptoms in PD patients. Therefore, this study holds significant implications for the management and treatment of PD.

This letter introduces a 3-D dual-polarized band-absorptive frequency selective rasorber (FSR) with a sharp pass-to-absorb transition property. The FSR with high selectivity is constructed by a composition made by a lossy layer and a band-stop layer. The working mechanism and the effectiveness of the FSR are analyzed through the equivalent circuit model. The 3-D unit cell is detailed, and a prototype is fabricated and measured in experiments. The proposed FSR possesses a pass-band below 3.02 GHz and an absorption band spanning from 5.37 GHz to 8.59 GHz, exhibiting a sharp spectral transition with a 1.78 FRap factor. The FSR is lightweight, mechanically robust and cost-effective, which is promising in stealthy radome applications for wideband UHF antennas.

A fast leaky-wave antenna (FLWA) with improved beam-scanning (BS) capability is proposed. The FLWA is made by a modified rectangular waveguide utilizing an array of metallic rods to enhance the spectral dispersion. In analysis, enhanced spectral dispersion is realized by mixing the Drude and Lorentz dispersions, which are determined by the waveguide and the rods. In simulation, the FLWA's BR rate is tailored by changing the metal rods, and enhanced spectral dispersion is realized below the cutoff of the original hollow waveguide. To realize impedance matching with the WR-90 waveguide, a gradient structure is designed. In experiments, a prototype is fabricated and measured. It is shown that the FLWA has a beam-scanning range from 12° to 72°, in the band from 7.15 GHz to 7.85 GHz. With these profiles, the BR rate is 6.42°/%, and an average gain is about 12.5 dBi in the whole operation band. The FLWA can be a good candidate for facilitating low-cost radars, including fast tracking, imaging, and communications.

Immunotherapy combined with chemotherapy regimen has been shown to be effective in recurrent or metastatic (R/M) head and neck squamous cell carcinoma (HNSCC). However, due to the small number of patients, its efficacy remains controversial in Asian populations, particularly in mainland China. Here a randomized, double-blind phase 3 trial evaluated the efficacy and safety of finotonlimab (SCT-I10A), a programmed cell death 1 (PD-1) monoclonal antibody, combined with cisplatin plus 5-fluorouracil (C5F) for the first-line treatment of R/M HNSCC. Eligible patients (n= 370) were randomly 2:1 assigned to receive finotonlimab plus C5F (n= 247) or placebo plus C5F (n= 123). The primary endpoint was overall survival (OS). In the finotonlimab plus C5F group, OS was 14.1 months (95% confidence interval (CI) 11.1–16.4), compared with 10.5 months (95% CI 8.1–11.8) in the placebo plus C5F group. The hazard ratio was 0.73 (95% CI 0.57–0.95, P= 0.0165), meeting the predefined superiority criteria for the primary endpoint. Finotonlimab plus C5F showed significant OS superiority compared with C5F alone and acceptable safety profile with R/M HNSCC, supporting its use as a first-line treatment option for R/M HNSCC. These results validate the efficacy and safety of the combination of finotonlimab and C5F in Asian patients with R/M HNSCC. ClinicalTrials.gov identifier: NCT04146402.

To improve the nonlinear fitting ability of multiple linear regression model in thermal error prediction of machine tools, an optimized thermal error modeling method based on test and simulation is proposed in this paper. Firstly, the finite-element analysis is used to complete the thermal-structure coupling analysis of the CNC machine tool. Based on the simulation results of temperature and deformation field of the machine tool, temperature-sensitive points (TSPs) are selected with fuzzy cluster analysis and correlation analysis. According to the simulation results, TSPs of the machine tool are fitted by quadratic polynomial from temperature to deformation. Finally, taking the deformation value of TSPs as the intermediate variable, the multiple linear regression model is established, and the optimized quadratic multiple regression thermal error model is obtained. The results show that the prediction curve of the optimized thermal error model based on on test and simulation is closer to the test curve, and the fitting index is better than that obtained from the traditional model. This method can effectively improve the nonlinear fitting ability of the thermal error model of CNC machine tools. Based on the thermal error model established in this paper, the compensation value of thermal error can be obtained by inputting the real-time measured temperature data of TSPs, which can effectively improve the machining accuracy of CNC machine tools.

Longitudinal magnetic resonance imaging (MRI) datasets have important implications for the study of degenerative diseases because such datasets have data from multiple points in time to track disease progression. However, longitudinal datasets are often incomplete due to unexpected quits of patients. In previous work, we proposed an augmentation method temporal recurrent generative adversarial network (TR-GAN) that can complement missing session data of MRI datasets. TR-GAN uses a simple U-Net as a generator, which limits its performance. Transformers have had great success in the research of computer vision and this article attempts to introduce it into longitudinal dataset completion tasks. The multihead attention mechanism in transformer has huge memory requirements, and it is difficult to train 3-D MRI data on graphics processing units (GPUs) with small memory. To build a memory-friendly transformer-based generator, we introduce a Hilbert transform module (HTM) to convert 3-D data to 2-D data that preserves locality fairly well. To make up for the insufficiency of convolutional neural network (CNN)-based models that are difficult to establish long-range dependencies, we propose an Swin transformer-based up/down sampling module (STU/STD) module that combines the Swin transformer module and CNN module to capture global and local information simultaneously. Extensive experiments show that our model can reduce mean squared error (MMSE) by at least 7.16% compared to the previous state-of-the-art method.

Stoichiometric mismatch between phytoplankton and zooplankton has implication for trophic transfer efficiency. Phosphorus (P) enrichment is expected to lower phytoplankton carbon (C) to P ratio (C : P) and thereby either alleviate P deficiency or induce excess P for zooplankton. However, the generality of zooplankton facing excess P and its effect on plankton trophic structure in natural systems are poorly understood. We analyzed the stoichiometry of seston and zooplankton, and plankton trophic structure in 32 (sub)tropical Chinese reservoirs. Our results showed that (1) stoichiometric mismatch between seston and zooplankton involved P or/and nitrogen (N) deficits in low‐nutrient reservoirs and P excess in high‐nutrient reservoirs; (2) at given seston C and phytoplankton compositional food quality levels, zooplankton to phytoplankton biomass ratio (Zoo : Phyto) showed a two‐segment piecewise relationship to seston N : P with a maximum at a breakpoint of 12.6 and strong reductions toward the extremes of seston N : P gradient; (3) increasing stoichiometric mismatch between seston and consumers reduced the contribution of cladocerans to zooplankton biomass and increased the trophic position of copepods; and (4) chlorophyll a(Chl a) to total phosphorus (TP) ratio (Chl a: TP) increased with decreasing Zoo : Phyto, and at a given Zoo : Phyto, it was higher in reservoirs with zooplankton dominated by copepods than in reservoirs with zooplankton dominated by cladocerans. These findings suggest that nutrient enrichment might improve stoichiometric constraint on plankton trophic structure in low‐nutrient reservoirs, but enhance negative effect of excess P in high‐nutrient reservoirs. Thus, negative impact of excess P on zooplankton may be a mechanism partly contributing to low Zoo : Phyto and high Chl a: TP in eutrophic reservoirs.

In recent years, controllable circularly polarized luminescence (CPL) materials with a high luminescence dissymmetry factor (glum) have attracted great interest due to their potential applications, such as optical display, optical data storage, optical encryption, and anticounterfeiting, which has become a popular research topic in the field of functional optoelectronic materials. In this work, by incorporating achiral CdSe/ZnS quantum dots (QDs) and the photochromic molecule, hydroxyl-containing spiropyran (SP–OH), into chiral nematic liquid crystals (N*-LCs) with a regular self-assembled helical superstructure, reversibly phototuning CPL in color and glumvalues based on the Förster resonance energy-transfer process from CdSe/ZnS QDs to SP–OH is successfully developed. The change of emission intensity and the enhancement of glumvalue demonstrate the existence of efficient energy transfer, which can be regulated on the basis of reversible photoisomerization process of SP–OH, resulting in various CPL emissions with enhanced glumvalue up to 1.0. This work can potentially be applied to chiral information storage and coding. Meanwhile, this provides a new method and idea for the preparation of intelligent and efficient CPL materials.

Ultrafast electron pulses have broad applications in the investigation of the ultrafast dynamics of materials and near-field nanophotonics. A hexagonal boron nitride (hBN) photoemission source has been proposed recently, characterized by a nanoscale working area and high brightness. However, the photoemission mechanisms of hBN are still not clear because the wide bandgap and low electron density of states of hBN are believed to result in low photoemission brightness. Here, we experimentally demonstrated the defect-assisted two-photon photoemission process in hBN by electron microscopy. The laser-induced defect states work as intermediate states to enhance the photoemission process by transferring a two-photon process to a cascaded one-photon process. In addition, we proposed another strategy to improve the photoemission brightness by increasing the density of states with a narrow-bandgap, two-dimensional material stacked on hBN. The photoemission intensity of the monolayer transition metal dichalcogenides (TMDs)/hBN heterostructure was largely enhanced, whereas the emission angle and energy spread remained similar to hBN. However, the photoemission intensity of TMDs/hBN can be influenced negatively by laser-induced defect trapping from the actual intermediate energy levels of TMDs. The defect-assisted photoemission process and heterostructure stacking strategy proposed here are instructive for the design of next-generation photoemission sources.

Rendering invisibility in the wide application scenarios has seen a surge in interest in recent years. Though various approaches have been proposed to realize concealments under different conditions, achieving polarization-independent invisibility for large objects remains a big challenge. Here, we propose to attain invisibility of a large dielectric slab with polarization constraints being totally lifted. This is accomplished by employing an antiscattering coating made of anisotropic metamaterials. We show that by tailoring the electric resonance of a triangular mushroom structure, antiphase electric dipole moment can be induced, resulting in an antipolarization response of the whole metamaterial coatings. By putting the proposed coatings on both sides of a large dielectric slab, a neutralization effect of the total polarization is observed, leading to the peculiar phenomenon of full-polarization invisibility. Our results are validated through full-wave simulations and experimental measurements. Remarkably, the intrinsic null-polarization property of the coating-slab-coating structure guarantees the invisibility feature of a large-scale bulk made by simply stacking the sandwiched composites, which facilitates the application of invisibility in practical scenarios such as the invisibility cloaks and the reflectionless antenna radomes.

Autonomous driving confronts great challenges in complex traffic scenarios, where the SOTIF risk can be triggered by the dynamic operational environment and system insufficiencies. The SOTIF risk is reflected not only intuitively in the collision risk with objects outside the autonomous vehicles, but also inherently in the performance limitation risk of the implemented algorithms. How to minimize the SOTIF risk for autonomous driving is currently a critical, difficult, and unresolved issue. Therefore, this paper proposes the “Self-Surveillance and Self-Adaption System” as a systematic approach to online minimize the SOTIF risk, which aims to provide a systematic solution for monitoring, quantification, and mitigation of inherent and external risks. As a demonstration of the system, the risk monitoring of the perception algorithm is highlighted. Moreover, the inherent perception algorithm risk and external collision risk are jointly quantified via SOTIF entropy, which is then propagated downstream to the decision-making module and mitigated. Finally, Hardware-in-the-Loop experiments are conducted to verify the efficiency and effectiveness of the system. The results demonstrate that the system enables dependable online monitoring, quantification, and mitigation of SOTIF risk in real-time critical traffic environments.

Spasticity is a motor disorder integrated in the upper motor neuron syndrome resulting from central nerve diseases such as stroke. The multifactorial nature of spasticity manifestations leads to the inter-rater and intrarater reliability of clinical assessment, hence, the objective severity quantification of the spastic hypertonia has attracted significant attention in the context of post-stroke rehabilitation. Here, we developed a novel assessment system to reliably identify the exaggerated muscle tone and quantitatively estimate the symptom severity in patients with upper limb spasticity. Twenty subjects with post-stroke spasticity (53.0 ± 13.9 years old) and ten age-matched healthy subjects performed the passive stretch movements under the single-task and dual-task protocols while wearing an exoskeletal measurement device developed by us. A preliminary identification layer was designed to discriminate the pathological electrophysiological outputs of the upper extremity muscles by using the long short-term memory (LSTM) networks. In the next layer, the severity quantification models can be triggered in parallel, aiming at evaluating the neural and non-neural level pathologies underlying the spastic resistance manually percepted by clinicians, where the muscle activation/co-activation features, kinematic departure, and biomechanical characteristics were considered to improve the clinical relevance. Based on these single-level decisions, the third layer was constructed as an integrated model to yield a more comprehensive quantification of the symptom severity. The experimental validation of the proposed system demonstrated good reliability in discriminating the spastic hypertonia from the normal muscle tone, as well as strong agreement of the quantitative severity estimations with the commonly accepted clinical scales for the neural level ${(R = 0.79,\;P = 2.79e - 5)}$ , non-neural level ${(R = 0.75,\;P = 1.62e - 4)}$ , and integrated level ${(R = 0.86,\;P = 9.86e - 7)}$ . In conclusion, the proposed assessment system holds great promise to provide clinicians with an easy-to-use tool as suitable support for spasticity diagnosis, disease monitoring, and treatment adjustment.

Graph convolutional network (GCN) outputs powerful representation by considering the structure information of the data to conduct representation learning, but its robustness is sensitive to the quality of both the feature matrix and the initial graph. In this article, we propose a novel multigraph fusion method to produce a high-quality graph and a low-dimensional space of original high-dimensional data for the GCN model. Specifically, the proposed method first extracts the common information and the complementary information among multiple local graphs to obtain a unified local graph, which is then fused with the global graph of the data to obtain the initial graph for the GCN model. As a result, the proposed method conducts the graph fusion process twice to simultaneously learn the low-dimensional space and the intrinsic graph structure of the data in a unified framework. Experimental results on real datasets demonstrated that our method outperformed the comparison methods in terms of classification tasks.

Amid the proliferation of immersive extended reality (XR), three key challenges are identified: the transmission and synthesis of intensive volumetric streams, the need for personalized XR experiences, and the availability of a comprehensive 3D data capturing system. To overcome these challenges, we propose a “Digitized You in My Eye” (DYME) spatial communication prototype, which leverages 5G, distributed computing, and artificial intelligence (AI) to address challenges in XR. DYME proposes a 3D live volumetric chat procedure that eliminates real-time image capture by utilizing pre-existing digitized 3D models stored in the cloud. It further embraces a distributed methodology for volumetric streaming, encompassing the utilization of cloud, edge, and local devices, capitalizing on the proximity of user infrastructure, and employing expeditious transmission facilitated by AI generative models. Additionally, a user-specific management module (DYME manager) is integrated to personalize XR experiences based on user behaviors and preferences. By combining these technologies, DYME aims to provide efficient, tailored, and immersive XR experiences. The prototype demonstrates the potential of AI and 5G-driven solutions in addressing challenges and enhancing user experiences in XR environments.

Contrastive learning has been successfully applied in unsupervised representation learning. However, the generalization ability of representation learning is limited by the fact that the loss of downstream tasks (e.g., classification) is rarely taken into account while designing contrastive methods. In this article, we propose a new contrastive-based unsupervised graph representation learning (UGRL) framework by 1) maximizing the mutual information (MI) between the semantic information and the structural information of the data and 2) designing three constraints to simultaneously consider the downstream tasks and the representation learning. As a result, our proposed method outputs robust low-dimensional representations. Experimental results on 11 public datasets demonstrate that our proposed method is superior over recent state-of-the-art methods in terms of different downstream tasks. Our code is available at https://github.com/LarryUESTC/GRLC.

Graphical Abstract:

Using lower limb rehabilitation robots (LLRRs) to help stroke patients recover their walking ability is attracting more and more attention presently. Previous studies have shown that gait rehabilitation training with natural gait pattern can improve the therapeutic outputs. However, how to generate the personalized gait trajectory has not been well researched. In this paper, a personalized gait generation method based anthropometric features is proposed. Firstly, gait trajectories are fitted and simplified into Fourier coefficient vectors, which are used to represent gait trajectories. Secondly, fourteen body features are used to generate the personalized gait trajectories and the feature set is further optimized based on the minimal redundancy maximal relevance criterion for easy application on the LLRR. Then, the relationship between the optimized feature set and gait trajectories is modeled by using the RF algorithm. Finally, the performance of the proposed method is demonstrated by several comparison experiments.

Recently, with the superior capacity for analyzing the multiplex graph data, self-supervised multiplex graph representation learning (SMGRL) has received much interest. However, existing SMGRL methods are still limited by the following issues: (i) they generally ignore the noisy information within each graph and the common information among different graphs, thus weakening the effectiveness of SMGRL, and (ii) they conduct negative sample encoding and complex pretext tasks for contrastive learning, thus weakening the efficiency of SMGRL. To solve these issues, in this work, we propose a new framework to conduct effective and efficient SMGRL. Specifically, the proposed method investigates the intra-graph and inter-graph decorrelation losses, respectively, for reducing the impact of noisy information within each graph and capturing the common information among different graphs, to achieve the effectiveness. Moreover, the proposed method does not need negative samples for the SMGRL and designs a simple pretext task, to achieve the efficiency. We further theoretically justify that our method achieves the maximal mutual information instead of directly conducting contrastive learning and theoretically justify that our method actually minimizes the multiplex graph information bottleneck, which guarantees the effectiveness. In addition, an extension for semi-supervised scenarios is proposed to fit the case that a few labels are provided in reality. Extensive experimental results verify the effectiveness and efficiency of the proposed method with respect to various downstream tasks.

Parkinson’s disease (PD) is a kind of neurological diseases. As one of its most common symptoms, bradykinesia mainly manifests as the slowing of fine movements of the upper limbs, and it seriously affects the quality of life. In this article, a temporal convolutional network (TCN) model-based Parkinson’s bradykinesia grading evaluation system is proposed. Firstly, a wearable device based on inertial sensors and low-energy Bluetooth was developed to collect kinematic information, 66 subjects from the Peking Union Medical College Hospital participated in the data collection work, and a dataset with four degrees of bradykinesia (Normal, Mild, Moderate, and Severe) was built. Afterward, a TCN model containing six residual block (RB) layers was designed for the evaluation of bradykinesia grades. The method showed better performance than the methods commonly used for bradykinesia detection in the comparison experiments. It provides a feasible method of PD bradykinesia grades assessment for assisting clinical diagnosis.

Alzheimer’s disease (AD) is an irreversible neurodegenerative disease, the most common form of dementia, affecting millions worldwide. Neuroimaging-based early AD diagnosis has become an effective approach, especially by using structural magnetic resonance imaging (sMRI). The convolutional neural network (CNN)-based method is challenging to learn dependencies between spatially distant positions in the various brain regions due to its local convolution operation. In contrast, the graph convolutional network (GCN)-based work can connect the brain regions to capture global information but is not sensitive to the local information in a single brain region. Unlike a separate CNN or GCN-based method, we proposed a brain-inspired global-local information fusion network (BGL-Net) to diagnose AD. It essentially inherits the advantages of both CNN and GCN. The experiments on three public data sets demonstrate the effectiveness and robustness of our BGL-Net. Our method achieved the best performance on three popular public data sets compared with the existing CNN and GCN-based methods. In addition, our visualization results of the learned brain connection on AD and normal people agree with many current AD clinical research.

The composite technology based on friction stir processing (FSP) can enhance the strength of Al alloys while retaining the original toughness and ductility of Al alloys, thus achieving an improvement in the overall performance of the composites. Ni/Al composites were prepared in three typical Al matrixes by optimized FSP, and the effect of in situ generated intermetallic compounds on the strengthening of different Al substrates during the FSP was investigated. It was found that the reinforcing phase of Ni/Al composites was mainly in situ generation of Al3Ni intermetallic compound using energy dispersive spectrometry (EDS) and transmission electron microscopy (TEM), and the more the number and more uniform the distribution of Al3Ni reinforcing phase, the better the mechanical properties of Ni/Al composites. The results indicate that the Ni/Al composites formed from different series of Al alloy matrix, the Ni/6061Al composite has the most remarkable microstructure uniformity, the greatest strength increase and the least elongation decrease, which is conducive to improving its strength without affecting the processing properties of the 6061 Al alloy. The finding opens new opportunities for the preparation of Ni/Al composites with excellent properties from 6061 Al alloys.

Lacking high‐performance nonlinear optical (NLO) materials limits the applications to opto‐electronic devices. In this work, a strategy of self‐assembling driven by π‐π stacking is proposed to design high‐performance NLO materials. Here, the third‐order NLO responses of two carbon‐based fuller‐rylenes dimers (Fuller‐PMIand Fuller‐PDI) are theoretically studied. The results mainly highlight the following points: (1) these two dimers are in good thermodynamic and kinetic stability, and the interaction energies of Fuller‐PMIand Fuller‐PDIdimers have reached to −63.50 kcal/mol and −68.18 kcal/mol, respectively; (2) the ratios of the static second order hyperpolarizability (γiiiiFF, i∈{x, y, z}) between the dimer and the monomer are γxxxxFFdimerγxxxxFFmonomer= 5.5, γyyyyFFdimerγyyyyFFmonomer= 1.2 and γzzzzFFdimerγzzzzFFmonomer= 1.2 for Fuller‐PMI, γxxxxFFdimerγxxxxFFmonomer= 1.1, γyyyyFFdimerγyyyyFFmonomer=1.4 and γzzzzFFdimerγzzzzFFmonomer= 4.2 for Fuller‐PDI. These results demonstrate that the self‐assembling driven by π‐π stacking is an effective strategy for designing high‐performance NLO materials. The self‐assembling strategy induced by π‐π stacking is utilized to design nonlinear optical (NLO) materials. Results suggest that this strategy is a very effective for designing high performance NLO materials. In this work, self‐assembly driven by π‐π stacking designs NLO materials with the remarkable second hyperpolarizability.

Achieving an omnidirectionally matched absorption in a broad frequency band is of particular interest in electromagnetic (EM) compatibility and shielding. Conventional absorbers used to work for a far-field incidence with a predefined direction, which do not show flexibility in practical applications involving various wave incidences. Here, we investigate the absorption performance of an inhomogeneous anisotropic slab for a given discrete source and show that an omnidirectionally matched absorption can be rendered, provided that the complex constitutive parameters of the slab obey a specific spatial profile. This specific profile guarantees that the wave impinging on each small grid of the slab will be absorbed without interface reflection, relying on the impedance matching nature of the Brewster’s effect. As a proof of concept, we implement a lumped-resistor-loaded gradient planar absorber, whose constitutive parameters fulfill the desired spatial profile in a broad frequency band. Both full-wave simulations and experimental measurements validate the results. The presented design methodology is simple, robust, and scalable to any other frequencies. Our work provides a novel solution to realize ideal EM shielding for discrete sources, which may be useful in practical scenes with given EM sources.

In this work, we focus on the challenging task, neuro-disease classification, using functional magnetic resonance imaging (fMRI). In population graph-based disease analysis, graph convolutional neural networks (GCNs) have achieved remarkable success. However, these achievements are inseparable from abundant labeled data and sensitive to spurious signals. To improve fMRI representation learning and classification under a label-efficient setting, we propose a novel and theory-driven self-supervised learning (SSL) framework on GCNs, namely Graph CCA for Temporal sElf-supervised learning on fMRI analysis (GATE). Concretely, it is demanding to design a suitable and effective SSL strategy to extract formation and robust features for fMRI. To this end, we investigate several new graph augmentation strategies from fMRI dynamic functional connectives (FC) for SSL training. Further, we leverage canonical-correlation analysis (CCA) on different temporal embeddings and present the theoretical implications. Consequently, this yields a novel two-step GCN learning procedure comprised of (i) SSL on an unlabeled fMRI population graph and (ii) fine-tuning on a small labeled fMRI dataset for a classification task. Our method is tested on two independent fMRI datasets, demonstrating superior performance on autism and dementia diagnosis. Our code is available at https://github.com/LarryUESTC/GATE.

Graph Convolutional Neural Networks (GCNs) are widely used for graph analysis. Specifically, in medical applications, GCNs can be used for disease prediction on a population graph, where graph nodes represent individuals and edges represent individual similarities. However, GCNs rely on a vast amount of data, which is challenging to collect for a single medical institution. In addition, a critical challenge that most medical institutions continue to face is addressing disease prediction in isolation with incomplete data information. To address these issues, Federated Learning (FL) allows isolated local institutions to collaboratively train a global model without data sharing. In this work, we propose a framework, FedNI, to leverage network inpainting and inter-institutional data via FL. Specifically, we first federatively train missing node and edge predictor using a graph generative adversarial network (GAN) to complete the missing information of local networks. Then we train a global GCN node classifier across institutions using a federated graph learning platform. The novel design enables us to build more accurate machine learning models by leveraging federated learning and also graph learning approaches. We demonstrate that our federated model outperforms local and baseline FL methods with significant margins on two public neuroimaging datasets.

Ketamine, a popular anesthetic, is often abused by people for its hallucinogenic effect. Thus, the safety of ketamine in pediatric populations has been called into question for potential neurotoxic effects. However, ketamine also has neuroprotective effects in many brain injury models. The differentiation of neural stem cells (NSCs) was influenced significantly by ketamine, but the molecular mechanism is still unclear. NSCs were extracted from the hippocampi of postnatal day 1 rats and treated with ketamine to induce NSCs differentiation. Our results found that ketamine promoted neuronal differentiation of NSCs dose-dependently in a small dose range (P < 0.001). The main types of neurons from NSCs were cholinergic (51 ± 4 %; 95 % CI: 41–61 %) and glutamatergic neurons (34 ± 3 %; 95 % CI: 27–42 %). Furthermore, we performed RNA sequencing to promise a more comprehensive understanding of the molecules regulated by ketamine. Finally, we combined bioimaging and multiple molecular biology techniques to clarify that ketamine influences NSC differentiation by regulating transient receptor potential canonical 3 (TRPC3) expressions. Ketamine dramatically repressed TRPC3 expression (MD [95 % CI]=0.67 [0.40–0.95], P < 0.001) with a significant increase of phosphorylated glycogen synthase kinase 3β (p-GSK3β; MD [95 % CI]=1.00 [0.74–1.27], P < 0.001) and a decrease of β-catenin protein expression (MD [95 % CI]=0.60 [0.32–0.89], P = 0.001), thereby promoting the differentiation of NSCs into neurons and inhibiting their differentiation into astrocytes. These results suggest that TRPC3 is necessary for ketamine to modulate NSC differentiation, which occurs partly via regulation of the GSK3β/β-catenin pathway.

With the rapid development of intelligent mining technology, remote-operated underground inspection and rescue robot have been widely used. This article recognized the operator’s emergency gestures based on forearm surface electromyography (sEMG). First, a wireless six-channel sEMG acquisition device is built and a dataset named coal mine inspection manipulator gestures (CMMG) is acquired; then, the features of each channel signal are extracted to a 2-D graph by a continue wavelet transform (CWT) method. The multistream convolutional neural network (CNN) model is built to analyze the feature graphs so as to detect action segments. The comparative experiments showed that the method improved the accuracy and showed better performance on both the CMMG dataset and the public Ninapro DB1 dataset.

The Jiama porphyry copper deposit in Tibet is one of the proven supergiant copper deposits in the Qinghai-Tibet Plateau at present, with the reserves of geological resources equivalent to nearly 20×106t. However, it features wavy and steep terrain, leading to extremely difficult field operation and heavy interference. This study attempts to determine the effects of the tensor controlled-source audio-magnetotellurics (CSAMT) with high-power orthogonal signal sources (also referred to as the high-power tensor CSAMT) when it is applied to the deep geophysical exploration in plateaus with complex terrain and mining areas with strong interference. The test results show that the high current provided by the high-power tensor CSAMT not only greatly improved the signal-to-noise ratio but also guaranteed that effective signals were received in the case of a long transmitter-receiver distance. Meanwhile, the tensor data better described the anisotropy of deep geologic bodies. In addition, the tests also show that when the transmitting current reaches 60 A, it is still guaranteed that strong enough signals can be received in the case of the transmitter-receiver distance of about 25 km, sounding curves show no near field effect, and effective exploration depth can reach 3 km. The 2D inversion results are roughly consistent with drilling results, indicating that the high-power tensor CSAMT can be used to achieve nearly actual characteristics of underground electrical structures. Therefore, this method has great potential for application in deep geophysical exploration in plateaus and mining areas with complex terrain and strong interference, respectively. This study not only serves as important guidance on the prospecting in the Qinghai-Tibet Plateau but also can be used as positive references for deep mineral exploration in other areas.

Most environment perception methods in autonomous vehicles rely on deep neural networks because of their impressive performance. However, neural networks have black-box characteristics in nature, which may lead to perception uncertainty and untrustworthy autonomous vehicles. Thus, this work proposes a decision-making method to adapt the potential perception uncertainty due to the sensor noises, fuzzy features, and unfamiliar inputs. The whole method is named as Perception Neural Networks Uncertainty Aware Decision-Making (PNNUAD) method. PNNUAD first uses the Monte Carlo dropout method to estimate the perception neural network uncertainty into a distribution around the original output. Then, the perception uncertainty will be considered in a designed reinforcement learning-based planner using a distributed value function. Finally, a backup policy will maintain the vehicle’s performance to avoid disastrous perception uncertainty. The evaluation section uses an augmented reality urban driving scenario; namely, the scenario builds in the CARLA simulator while the perception uncertainty comes from the real dataset. This case study focuses on the object class uncertainty of a widely used neural network, i.e., YOLO-V3. The results indicate that the proposed method can maintain AV safety even with poor perception performance. Meanwhile, the AV has not become too conservative by defending the perception uncertainty. This work is necessary for applying the statistics neural networks to safety-critical autonomous vehicles, and the source code will be open-source in this work.

Parkinson’s disease (PD) is the second most common neurodegenerative disorder and the heterogeneity of early PD leads to interrater and intrarater variability in observation-based clinical assessment. Thus, objective monitoring of PD-induced motor abnormalities has attracted significant attention to manage disease progression. Here, we proposed a hierarchical architecture to reliably detect abnormal characteristics and comprehensively quantify the multisymptom severity in patients with PD. A novel wearable device was designed to measure motor features in 15 PD patients and 15 age-matched healthy subjects, while performing five types of motor tasks. The abnormality classes of multimodal measurements were recognized by hidden Markov models (HMMs) in the first layer of the proposed architecture, aiming at motivating the evaluation of specific motor manifestations. Subsequently, in the second layer, three single-symptom models differentiated PD motor characteristics from normal motion patterns and quantified the severity of cardinal PD symptoms in parallel. In order to further analyze the disease status, the multilevel severity quantification was fused in the third layer, where machine learning algorithms were adopted to develop a multisymptom severity score. The experimental results demonstrated that the quantification of three cardinal symptoms was highly accurate to distinguish PD patients from healthy controls. Furthermore, strong correlations were observed between the Unified PD Rating Scale (UPDRS) scores and the predicted subscores for tremor ${(R = 0.75,\;P = 1.40e - 3)}$ , bradykinesia ${(R = 0.71,\;P = 2.80e - 3)}$ , and coordination impairments ${(R = 0.69,\;P = 4.20e - 3)}$ , and the correlation coefficient can be enhanced to ${0.88}\,\,{(P = 1.26e - 5)}$ based on the fusion schemes. In conclusion, the proposed assessment architecture holds great promise to push forward the in-home monitoring of clinical manifestations, thus enabling the self-assessment of disease progression.