Current learning-based stereo matching is generally poor in adaptively exploring the robust and salient features at different scenes, leading to ambiguity of matching, especially in challenging areas. To tackle this problem, inspired by the global representation of the graph, we propose a Graph Channel Attention (GCA) to globally and interactively learn binocular attention for robust stereo matching, instead of traditional separate local monocular attention. We first construct a 2D binocular graph structure with left and right subgraphs, where the left and right channel information can globally interact. After that, our interactive graph inference with cross interaction and inner aggregation is proposed to improve the linkage inference between and within binocular graphs, which can consider global and interactive attention information like real human eyes. Thus, our GCA alters the channel attention from traditional 1D to binocular 2D, which can imitate the global interaction and attention ability of real human eyes. Finally, we utilize the GCA into stereo matching, and experiment results show that our method demonstrates state-of-the-art performance on KITTI 2012/2015 and Middlebury Stereo Evaluation v.3.

Currently, deep learning-based unsupervised stereo matching approaches predominantly employ MLP-based convolutional networks. However, the inherent local receptive field of convolutional networks lacks sufficient global context, and the limited nonlinear modeling capacity of MLPs with fixed, non-learnable activations make unsupervised stereo matching struggle to accurately handle complex nonlinearity, noise, and ambiguity through multiple image-structure constraints, especially in challenging regions. To address these challenges, we propose a Multi-resolution and Multi-scale Graph Attention Network based on Kolmogorov-Arnold Networks (KAN), MMGA-KAN Net, for unsupervised stereo matching. First, multi-scale graph structure with global context inference is constructed, where the multi-scale nodes with projected multi-scale image features can be linked by message passing and cross-scale information interaction, instead of the inherent locality of convolutional networks and the limitations of single-scale graph structure. Second, we introduce MGA-KAN with KAN-based multi-scale graph structure and dynamic attention mechanism, which replaces the fixed, non-learnable activation functions and simple linear transformations of traditional MLP with the flexible, learnable B-spline functions by KAN. Then, KAN-based dynamic attention mechanism compatible with both global and local information overcomes limitations of traditional graph structure in capturing local details during global propagation. Finally, to mitigate noise accumulation and detail loss by global propagation of the graph structure at single resolution, we propose inner-graph inference and cross-image interaction with multi-resolution and multi-scale, which aggregate multi-scale information within the graph structure at identical resolution and update cross-resolution information across resolutions. Experiments demonstrate that our approach achieves state-of-the-art performance on Scene Flow, KITTI 2012/2015, and Middlebury Stereo Evaluation v.3/2021.

For lithium-ion cell health diagnosis, machine learning techniques have been widely used but still leave something to be desired. Specifically, Gaussian process regression (GPR) suffers from the exponential increase in computational load when training large sample data, which is not suitable for online deployment. In this study, the fuzzy information granulation (FIG) technique is first combined with the GPR model for short-term look- ahead state of health (SOH) estimation and long-term remaining useful life (RUL) prediction. The FIG technique divides the original data into fuzzy particles, which eliminates its volatility and uncertainty to improve prediction accuracy and reduce computational complexity. In the experiment, the laboratory data set and three public datasets under fast charging and constant current and constant voltage conditions are used to verify the effectiveness and robustness. Compared with the single GPR, the proposed method improves the prediction accuracy by 73.04 %, and the corresponding calculation time is reduced from 34.33s to 0.86s.

It is very crucial to accurately estimate the state-of-charge (SOC) and state-of-health (SOH) of electric vehicles. Considering that the ordinary least square method and Kalman filter have low data utilization and poor tracking ability, this research put forward a novel co-estimator on the ground of the multi-innovations (MI) principle. In this method, the parameters are calculated by forgetting factor MI least squares, SOC is estimated by the MI unscented Kalman filter, and the SOH is predicted by the extended Kalman filter. The proposed method is confirmed under the urban dynamometer driving schedule condition and the dynamic stress test condition at different temperatures. In the co-estimation, the maximum absolute error and root-mean-square error of SOC are only 0.53% and 0.3% respectively, 0.025% and 0.00852% respectively for SOH when the estimated effect is optimal. Under multiple test cycles, the estimated accuracy of SOH can also remain within 2%, but is slightly higher than that of SOC. The results also indicate that the proposed method has high precision and robustness in extreme environment.

AIMS:Health inequalities studies need to understand how individuals simultaneously defined by several socioeconomic factors differ from others when facing a series of stressors across the lifespan in the risk of major depression (MD). Theoretical efforts, as well as empirical studies, have suggested a pertinent role of social support in mental health outcomes. However, little is known about which forms of social support would alleviate the negative impact of MD vulnerability in self-rated mental health (SRMH) across different socioeconomic groups. We investigated 1) differential associations between lifetime stressors and MD across social groups and 2) explored diverse social support forms mediating the associations between MD vulnerability and SRMH. METHODS:Data analyzed were from a large longitudinal population-based cohort. Multilevel analysis of individual heterogeneity and discriminatory accuracy was used to articulate MD vulnerability in different social groups defined by ethnicity, gender, and socioeconomic status (SES). Genetic predispositions were also included in the modeling process. These social groups were then regrouped based on their vulnerability level of MD. Mediation analyses were then applied to identify which social support forms mediate the effect of MD vulnerability on SRMH. RESULTS:Higher levels of stressors were associated with higher risks of MD, and their associations varied by different social groups. The social groups (White men with medium SES or White women with high SES) had the lowest predicted incidence of MD, whereas White women with low SES reported the highest predicted incidence of MD. Two social support forms (guidance and opportunity for nurturance) significantly mediated the indirect paths between MD vulnerability and SRMH. CONCLUSIONS:By applying an intersectional lens, the present study provides a novel quantitative instrument for documenting the associations of stress and depression in various social identities. The findings of the study suggest more focused intervention programs and strategies for risk reduction should focus on identified characteristics and pay particular attention to the combined effect of lifetime stressors and discovered social identities.

To address the insufficient detection accuracy for steel strip surface defects, this paper proposes an improved YOLOv8n-based detection model. During the data preprocessing stage, multiple augmentation methods are employed to enhance dataset diversity, significantly improving the model's generalization capability. The architecture integrates a C2f-GAM feature extraction module that combines the GAM attention mechanism with the C2f module, effectively reducing feature information loss while lowering computational complexity. Furthermore, the feature pyramid network is reconstructed using GSConv lightweight convolution and VoV-GSCSP modules from the Slimneck architecture, substantially enhancing information retention capacity. The detection accuracy and convergence speed are further optimized through the adoption of the SIoU loss function. Experimental results demonstrate that the improved algorithm achieves 75.4% mAP@0.5 on the X-SDD dataset, with model parameters of 3.65M and computational complexity of 8.5 GFLOPS. Compared to the baseline model without data augmentation, this represents a 3.6% performance improvement while maintaining real-time processing capability. These results validate the effectiveness of the proposed method for steel strip surface defect detection in industrial applications.

Bisphenol A (BPA), as a representative endocrine disrupting compound, is harmful to human health even at very low doses. The chemiluminescence (CL) method used to detect BPA often uses hydrogen peroxide as an oxidant. Due to the instability of hydrogen peroxide, it is difficult to ensure the accuracy of the detection. Herein, a novel CL system based on disulfide bonds-containing perylene diimide-based conjugated polymer (SHPDI) and peroxymonosulfate (PMS) was designed for the rapid, sensitive and accurate detection of BPA. In this system, reactive oxygen species (ROS) generated by Co2+ and PMS activation oxidizes the S-S and -SH bonds in SHPDI to sulfoxide bonds, resulting in a strong CL signal. Meanwhile, the C--O bonds and delocalized pi electrons in SHPDI also promote PMS decomposition and the production of more ROS, which is beneficial for strong CL. The proposed CL sensor based on the SHPDI-Co(II)-PMS system exhibited a rapid and sensitive response to BPA in as short as 10 seconds, leading to a noticeable linear reduction in the CL intensity in the range of 0.005-2 mu M with a detection limit of 0.72 nM (S/N = 3). The CL method has been successfully used for the direct detection of trace amounts of BPA in barreled drinking water.

Thanks to the advantages of near-zero background signal, high sensitivity and wide dynamic range, electrochemiluminescence (ECL) has been widely used in various fields of biological analysis and clinical diagnosis. ECL probes are the key factors for ECL methods and directly determine the strategies and efficiency of applications. Therefore, exploring new and efficient ECL probes has garnered significant attention. As the theoretical basis and booster for the development of ECL probes with excellent analysis performance, the ECL mechanism has also received much attention. Herein, we review the ECL probes including inorganic metal complexes, organic molecules, and nanomaterials. Their related annihilation and/or coreactant ECL emitting mechanisms are summarized in combination with applications. Finally, the future development of ECL probes is prospected.

INTRODUCTION:Prior research has documented the associations among phubbing, depression, and anxiety, while the cross-sectional design failed to clarify the temporal directionality of the relationships between these mental disorders and behavioral issues. To bridge this gap, the present study utilizing longitudinal data aimed to articulate the temporal relationships between these mental disorders and behavioral issues. METHODS:A total of 3296 adolescents from China (54.5% girls; Mage = 15.17) participated in the study. Symptoms of phubbing, depression, and anxiety were assessed 18 months later (May 2023) after the baseline (November, 2021). The cross-sectional network and cross-lagged panel network models were conducted to explore the associations between the network structures of phubbing, depression, and anxiety. The network comparison test (NCT) was then performed to unveil whether the network structures vary based on school grade. RESULTS:In the cross-sectional network, significant differences in the overall structures between middle and high school students were observed. For the longitudinal network, the core symptoms responsible for temporal relationships were mostly between depressive and anxiety symptoms. Phubbing-related symptoms and restlessness (anxiety symptom) were the bridge symptoms of phubbing, depression, and anxiety. Besides, the central bridges associated with phubbing-related symptoms differed significantly across different school stages. CONCLUSIONS:Successfully regulating negative emotions can play a pivotal role in tackling the root causes linked to phubbing. Apart from addressing restlessness, future interventions focusing on nomophobia and interpersonal conflict in middle school students, as well as self-isolation in high school students, contributed to mitigating phubbing, depression, and anxiety.

BACKGROUND AND AIMS:Previous studies are limited in addressing the directionality of temporal relationships between problematic gaming and phubbing symptoms by exploring cross-sectional studies. Therefore, we estimated the longitudinal relationships between individual behavioral addictive symptoms including problematic gaming and phubbing in adolescence, and explored potential sex differences in these relationships. METHODS:This study included 3296 participants in Shandong Province, China. Data were collected from November 2021 (mean [SD] age: 15.17 [1.44] years) to May 2023 (mean [SD] age: 17.50 [1.18] years), with females comprising 54.5 % of the sample. Problematic gaming and phubbing were assessed using validated scales at each wave. We construct cross-sectional networks and cross-lagged panel networks (CLPN) to explore the contemptuous and temporal relationships between problematic gaming and phubbing. RESULTS:Contemporaneous networks revealed significant differences in problematic gaming and phubbing networks between males and females. Additionally, temporal network analyses indicated that among male adolescents, feeling anxious when unable to play games was the most influential predictor of subsequent behavioral addictive symptoms. For female adolescents, fantasizing about gaming had the most significant associations with future addictive behaviors. The strongest bridge symptom linking problematic gaming and phubbing for both sexes was focusing on phones rather than engaging in conversation. DISCUSSION AND CONCLUSIONS:The study applied network modeling to panel data from a large, population-based cohort of adolescents, identifying unique longitudinal relationships between problematic gaming and phubbing across symptom domains. It provides valuable insights into the characterization of behavioral addictive symptoms among adolescents and the potential predictive relationships among these symptoms among different sexes, guiding sex-specific targeted interventions for adolescents.

In light of deep tissue penetration and ultralow background, near-infrared (NIR) persistent luminescence (PersL) bioprobes have become powerful tools for bioapplications. However, the inhomogeneous signal attenuation may significantly limit its application for precise biosensing owing to tissue absorption and scattering. In this work, a PersL lifetime-based nanoplatform via deep learning was proposed for high-fidelity bioimaging and biosensing in vivo. The persistent luminescence imaging network (PLI-Net), which consisted of a 3D-deep convolutional neural network (3D-CNN) and the PersL imaging system, was logically constructed to accurately extract the lifetime feature from the profile of PersL intensity-based decay images. Significantly, the NIR PersL nanomaterials represented by Zn1+xGa2-2xSnxO4: 0.4 % Cr (ZGSO) were precisely adjusted over their lifetime, enabling the PersL lifetime-based imaging with high-contrast signals. Inspired by the adjustable and reliable PersL lifetime imaging of ZGSO NPs, a proof-of-concept PersL nanoplatform was further developed and showed exceptional analytical performance for hypochlorite detection via a luminescence resonance energy transfer process. Remarkably, on the merits of the dependable and anti-interference PersL lifetimes, this PersL lifetime-based nanoprobe provided highly sensitive and accurate imaging of both endogenous and exogenous hypochlorite. This breakthrough opened up a new way for the development of high-fidelity biosensing in complex matrix systems.

Labile toxic pollutants detection remains a challenge due to the problem that a single method is prone to producing false-negative/-positive signals. The construction of a multisignal sensing platform with the advantages of different strategies is an effective way to solve this problem. Herein, a novel resonant light scattering (RLS), fluorescent and rapid visual multisignals sensing strategy for p-aminophenol (p-AP) detection was designed based on the adsorption and oxidation effects of defective amino-functionalized Ag-based nano metal-organic frameworks (NH2-Ag-nMOFs). In this reaction process, NH2-Ag-nMOFs with incomplete coordination oxidize H2O2 to produce singlet oxygen (O-1(2)) which rapidly oxidizes p-AP, leading to the reduction of Ag+ to Ag-0, thereby disrupting the structure of NH2-Ag-nMOFs and resulting in fluorescence quenching of NH2-Ag-nMOFs. Synchronously, owing to Ag-0 aggregation and p-AP oxidation, the color of the system changed from colorless to purplish-red and pale brown within 20 s. The assay has realized the rapid naked-eye detection of 5 mu M p-AP rapidly. Additionally, thanks to the intermolecular hydrogen bonding, NH2-Ag-nMOFs-p-AP aggregates formed, which enhanced the RLS signal. With the RLS signal, the designed multisignals sensing platform can analyze p-AP at a concentration as low as 11 nM and yield a wider dynamic response range than any single signal strategy reported before, which can quickly meet the measurement requirement of different actual samples. Overall, the proposed strategy without assembling various signal indicators presented an accurate, rapid, cost-effective, and sensitive multisignals sensing platform for p-AP analysis and has great prospects in labile toxic pollutants monitoring.

化学发光法因其灵敏度高、选择性好、响应速度快等优异性能在分析传感领域引起广泛关注.然而目前发展发光性能优异且发射波长长的新型化学发光体系仍面临严峻挑战.本文通过将具有优异反应活性和近红外光学发射信号的铬离子掺杂镓酸锌(Cr 3+ -doped ZGC)纳米材料引入新型化学发光体的构建中,发展了基于ZGC-Fe 2+ -H 2 O 2 的新型近红外化学发光体系,成功用于H 2 O 2 的高效传感研究.更进一步地,通过对ZGC-Fe 2+ -H 2 O 2 体系的化学发光光谱及活性氧自由基清除实验研究,提出了一种可能的基于活性氧注入式的“电子-空穴复合”发光机制.得益于化学发光信号强度高且免背景信号干扰的独特优势,该ZGC-Fe 2+ -H 2 O 2 近红外化学发光法在检测H 2 O 2 时具有宽线性范围（10-100μM）,高灵敏度、检测限低至4.68μM以及优异的抗干扰性和选择性,为其他化学发光检测平台的构建提供了新思路.

Abstract Background Stressors across the lifespan are associated with the onset of major depressive disorder (MDD) and increased severity of depressive symptoms. However, it is unclear how lifetime stressors are related to specific MDD subtypes. The present study aims to examine the relationships between MDD subtypes and stressors experienced across the lifespan while considering potential confounders. Methods Data analyzed were from the Zone d’Épidémiologie Psychiatrique du Sud-Ouest de Montréal (N = 1351). Lifetime stressors included childhood maltreatment, child–parent bonding, and stressful life events. Person-centered analyses were used to identify the clusters/profiles of the studied variables and multinomial logistic regression analyses were performed to examine the relationships between stressors and identified MDD subtypes. Intersectional analysis was applied to further examine how distal stressors interact with proximal stressors to impact the development of MDD subtypes. Results There was a significant association between proximal stressors and melancholic depression, whereas severe atypical depression and moderate depression were only associated with some domains of stressful life events. Additionally, those with severe atypical depression and melancholic depression were more likely to be exposed to distal stressors such as childhood maltreatment. The combinations of distal and proximal stressors predicted a greater risk of all MDD subtypes except for moderate atypical depression. Conclusions MDD was characterized into four subtypes based on depressive symptoms and severity. Different stressor profiles were linked with various MDD subtypes. More specific interventions and clinical management are called to provide precision treatment for MDD patients with unique stressor profiles and MDD subtypes.

Background: Both genetic predispositions and exposures to stressors have collectively contributed to the development of major depressive disorder (MDD). To deep dive into their roles in MDD, our study aimed to examine which susceptible gene expression interacts with various dimensions of stressors in the MDD risk among a large population cohort. Methods: Data analyzed were from a longitudinal community-based cohort from Southwest Montreal, Canada (N = 1083). Latent profile models were used to identify distinct patterns of stressors for the study cohort. A transcriptome-wide association study (TWAS) method was performed to examine the interactive effects of three dimensions of stressors (threat, deprivation, and cumulative lifetime stress) and gene expression on the MDD risk in a total of 48 tissues from GTEx. Additional analyses were also conducted to further explore and specify these associations including colocalization, and fine-mapping analyses, in addition to enrichment analysis investigations based on TWAS. Results: We identified 3321 genes linked to MDD at the nominal p-value <0.05 and found that different patterns of stressors can amplify the genetic susceptibility to MDD. We also observed specific genes and pathways that interacted with deprivation and cumulative lifetime stressors, particularly in specific brain tissues including basal ganglia, prefrontal cortex, brain amygdala, brain cerebellum, brain cortex, and the whole blood. Colocalization analysis also identified these genes as having a high probability of sharing MDD causal variants. Limitations: The study cohort was composed exclusively of individuals of Caucasians, which restricts the generalizability of the findings to other ethnic population groups. Conclusions: The findings of the study unveiled significant interactions between potential tissue-specific gene expression x stressors in the MDD risk and shed light on the intricate etiological attributes of gene expression and specific stressors across the lifespan in MDD. These genetic and environmental attributes in MDD corroborate the vulnerability-stress theory and direct future stress research to have a closer examination of genetic predisposition and potential involvements of omics studies to specify the intricate relationships between genes and stressful environments.

To the deep tissue penetration and ultra-low background, developing near-infrared (NIR) chemiluminescence probes for human health and environmental safety has attracted more and more attention, but it remains a huge challenge. Herein, a novel NIR chemiluminescence (CL) system was rationally designed and developed, utilizing Cr3+-activated ZnGa2O4 (ZGC) nanoparticles as a catalytic luminophore via hypochlorite (NaClO) activation for poisonous target (hydrazine, N2H4) detection. With superior optical performance and unique catalytic structure of ZGC nanoparticles, the fabricated ZGC-NaClO-N2H4 CL system successfully demonstrated excellent NIR emission centered at 700 nm, fast response, and high sensibility (limit of detection down to 0.0126 μM). Further experimental studies and theoretical calculations found the cooperative catalytic chemiluminescence resonance energy transfer mechanism in the ZGC-NaClO-N2H4 system. Remarkably, the ZGC-based NIR CL system was further employed for N2H4 detection in a complicated matrix involving bioimaging and real water samples, thereby opening a new way as a highly reliable and accurate tool in biomedical and environmental monitoring applications.

To address challenges in steel surface defect detection, such as low accuracy and slow processing speed, an enhanced algorithm is proposed. The C3 module is replaced with GSConv (multi-channel shuffle convolution) to improve parameter efficiency and detection accuracy. The nearest neighbor up-sampling is substituted with CARAFE (a lightweight operator) to enhance the receptive field and utilize semantic information. GhostNetV2 replaces the CBS convolutional module for efficient feature extraction through stacking operations. Additionally, the Hard Swish activation function is incorporated to boost accuracy. Experimental results on the NEU-DET dataset demonstrate that the average accuracy of the improved YOLOv5s algorithm is 81.2%, a 1.7% increase compared to the original algorithm, the amount of computation is 17.2GFLOPs and the number of model parameters is 8.65M. The model effectively enhances detection accuracy.

The current diagnostic criteria for depression do not sufficiently reflect its heterogeneous clinical presentations. Associations between adverse childhood experiences (ACEs), allostatic load (AL), and depression subtypes have not been extensively studied. Depression subtypes were determined based on clinical presentations, and their relationships to AL biomarkers and ACEs were elucidated in a sample of middle-aged and older adults. Participants from the Canadian Longitudinal Study on Aging who screened positive for depression were included (n=3966). Depression subtypes, AL profiles and ACE profiles were determined with latent profile analyses, and associations between them were determined using multinomial logistic regression. Four depression subtypes were identified: positive affect, melancholic, typical, and atypical. Distinct associations between depression subtypes, stressor profiles and covariates were observed. Among the subtypes compared to positive affect, atypical subtype had the most numerous significant associations, and the subtypes had unique relationships to stressor profiles. Age, sex, smoking status, chronic conditions, marital status, and physical activity were significant covariates. The present study describes distinct associations between depression subtypes and measures of stress (objective and self-reported), as well as related factors that differentiate subtypes. The findings may inform more targeted and integrated clinical management strategies for depression in individuals exposed to multiple stressors.

Although there is a growing awareness of the co-occurrence of internet gaming disorder (IGD) with other mental health problems, the specific patterns of how these symptoms interact over time, especially across different age groups, remain unclear. The current study utilizes cross-lagged panel network modeling (CLPN) to investigate the dynamic, longitudinal relationships among symptoms of IGD, depression and anxiety among adolescents across time, and how these connections change with different developmental stages. A total of 3296 middle and late adolescents who have finished 3-time points research were included in the present study. Significant differences were found between middle and late adolescents in the structures and strengths of the contemporaneous and longitudinal networks. For middle adolescents, symptoms tended to predict subsequent symptoms within the same disorder. However, late adolescents showed a stronger trend of symptoms being interconnected across comorbid conditions. Feelings of worthlessness & hopelessness were the most impactful symptoms for middle adolescents in the short term and they continued to significantly affect late adolescents in the long term. In addition, restless and suicide or self-harm were the most important bridge symptoms for middle and late adolescents, respectively. This study emphasizes the importance of developing targeted intervention strategies focusing on both central and bridging symptoms of the comorbid conditions of IGD, depression, and anxiety in adolescence. Recognizing distinct adolescents' needs, interventions should be tailored to effectively address the unique challenges at different developmental stages.

BACKGROUND:This study aimed to investigate the symptom patterns of major depressive disorder (MDD) and generalized anxiety disorder (GAD) in a matched nationally representative sample of the Canadian population. We also tested whether childhood maltreatment (CM) exposures and sex would be linked with different symptom patterns.METHODS:A total of 3296 participants from the Canadian Community Health Survey-Mental Health with complete information on MDD and GAD symptoms and being matched on the studied sociodemographic characteristics were included in the current study. Network analysis was performed to examine the MDD-GAD symptom network, network stability and centrality indices were also estimated. Finally, network comparison in connectivity patterns was conducted to explore the impact of maltreatment experience and sex differences in the MDD-GAD symptom networks.RESULTS:The CM group had stronger network connections and showed differences in the network structures from the non-CM group. In the CM group, depressed mood and diminished interest were central symptoms and strongly connected with other symptoms. Additionally, females had stronger connections in the MDD-GAD symptom network than males, and sleep disturbance was a central symptom for females, alongside depressed mood and diminished interest.LIMITATIONS:The cross-sectional design restricts our capacity to establish longitudinal or causal connections between symptoms.CONCLUSIONS:Depressed mood was the most central node that was strongly connected with other symptoms in the network. Distinct MDD-GAD symptom networks were discovered in the CM and the female group when compared to their counterparts. Noteworthy, individuals with CM had a stronger correlation between worry and suicidal ideation. Clinical management and intervention efforts should pay close attention to these core symptoms to yield optimal treatment effects, particularly for females and individuals with CM.