Search

Abstract Detecting taxi passengers is crucial for assessing taxi driver behavior, which plays a significant role in regulating the taxi industry. Despite the advancements in deep learning, object detection algorithms have not been extensively applied to this domain. In this article, an innovative taxi passenger detection algorithm is introduced based on YOLOv8, a lightweight and highly accurate method designed to automatically monitor driver behavior and regulate the taxi industry. To address the challenge of deploying complex object detection models on mobile devices, the ghost module is incorporated in place of standard convolutions within the C2f module, thereby making the model more lightweight. Furthermore, the model's performance is enhanced by integrating an improved version of Frequency Channel Attention (FCA), termed Parallel Frequency Channel Attention (PFCA), which boosts detection accuracy with minimal additional parameters and computational overhead. Experimental results on a specific taxi passenger dataset demonstrate that the proposed method significantly outperforms the baseline YOLOv8n model. Specifically, the model reduces the number of parameters and floating point operations by 12.96% and 8.18%, respectively, while achieving increases in mAP50 and mAP50‐95 by 0.27 and 0.73 percentage points, respectively.

Abstract Thallium, a highly toxic pollutant, shows greater toxicity to human than other common heavy metals such as mercury, lead, cadmium and its effective removal from wastewater gains great attention. The main restriction for the Tl+ removal is the interference of a high concentration of co-existing ions in wastewater. Therefore, the goal of the current work was to synthesis adsorbent with high selectivity for the Tl+ removal. Herein, the pore size sieving strategy was proposed and Prussian blue-impregnated biochar (BC@PB) particles was synthesized. More than 95% Tl+ can be removed even the concentrations of the coexistence ions (Na+, Cd2+, and Zn2+) 1,000 higher than the initial concentration of Tl+ (500 μg/L). BC@PB also showed large adsorption capacity (9365 μg/g) and more than 99% Tl+ (initial concentration, 500 μg/L) were removed in just 1 min. The BC@PB had excellent and stable Tl+ removal ability (> 99%) over a range of pH from 3 to 9, which covered the pH range of common thallium-containing wastewater. The density functional theory (DFT) calculation confirmed that not only hydrated volume but also the hydration free energy of ions, which governed the energy barrier for ions entering into narrow channels of BC@PB, played essential roles on the selectivity removal of Tl+. Overall, due to its high selectivity, high adsorption capacity and easy preparation process, the synthesized BC@PB particles based on the pore sizing sieving strategy, can be a promising candidate for the removal of thallium from wastewater.

Multimorbidity accelerates cognitive decline and leads to an increased risk of cognitive impairment. However, existing studies have mainly explored the cognitive status of patients with a single or specific chronic disease, and the patient with multimorbidity remains to be urgently explored. The present study describes the epidemiological characteristics of cognitive impairment in multimorbidity, summarizes the influencing factors, organizes the association patterns between multimorbidity and cognitive impairment, elucidates the mechanisms underlying their occurrence, and finally proposes preventive and control strategies. The findings of this study are intended to serve as a valuable reference for future efforts in preventing and treating cognitive impairment in multimorbidity.

Abstract Quantum chemical (QC) property prediction is crucial for computational materials and drug design, but relies on expensive electronic structure calculations like density functional theory (DFT). Recent deep learning methods accelerate this process using 1D SMILES or 2D graphs as inputs but struggle to achieve high accuracy as most QC properties depend on refined 3D molecular equilibrium conformations. We introduce Uni-Mol+, a deep learning approach that leverages 3D conformations for accurate QC property prediction. Uni-Mol+ first generates a raw 3D conformation using RDKit then iteratively refines it towards DFT equilibrium conformation using neural networks, which is finally used to predict the QC properties. To effectively learn this conformation update process, we introduce a two-track Transformer model backbone and a novel training approach. Our benchmarking results demonstrate that the proposed Uni-Mol+ significantly improves the accuracy of QC property prediction in various datasets.

Abstract The relationship between green and grey urban infrastructure, local meteorological conditions, and traffic-related air pollution is complex and dynamic. This case study examined the effect of evolving morphologies around a city square park in Dublin and explores the twin impacts of local urban development (grey) and maturing parks (green) on particulate matter (PM) pollution. A fixed air quality monitoring campaign and computational fluid dynamic modelling (ENVI-met) were used to assess current (baseline) and future scenarios. The baseline results presented the distribution of PM in the study area, with bimodal (PM2.5) and unimodal (PM10) diurnal profiles. The optimal vegetation height for air quality within the park also differed by wind direction with 21 m vegetation optimal for parallel winds (10.45% reduction) and 7 m vegetation optimal for perpendicular winds (30.36% reduction). Increased building heights led to higher PM2.5 concentrations on both footpaths ranging from 25.3 to 37.0% under perpendicular winds, whilst increasing the height of leeward buildings increased PM2.5 concentrations by up to 30.9% under parallel winds. The findings from this study provide evidence of the importance of more in-depth analysis of green and grey urban infrastructure in the urban planning decision-making process to avoid deteriorating air quality conditions around city square parks.

Abstract Despite being one of the most prevalent RNA modifications, the role of N6‐methyladenosine (m6A) in amyotrophic lateral sclerosis (ALS) remains ambiguous. In this investigation, we explore the contribution of genetic defects of m6A‐related genes to ALS pathogenesis. We scrutinized the mutation landscape of m6A genes through a comprehensive analysis of whole‐exome sequencing cohorts, encompassing 508 ALS patients and 1660 population‐matched controls. Our findings reveal a noteworthy enrichment of RNA binding motif protein X‐linked (RBMX) variants among ALS patients, with a significant correlation between pathogenic m6A variants and adverse clinical outcomes. Furthermore, Rbmx knockdown in NSC‐34 cells overexpressing mutant TDP43Q331K results in cell death mediated by an augmented p53 response. Similarly, RBMX knockdown in ALS motor neurons derived from induced pluripotent stem cells (iPSCs) manifests morphological defects and activation of the p53 pathway. Transcriptional analysis using publicly available single‐cell sequencing data from the primary motor cortex indicates that RBMX‐regulated genes selectively influence excitatory neurons and exhibit enrichment in ALS‐implicated pathways. Through integrated analyses, our study underscores the emerging roles played by RBMX in ALS, suggesting a potential nexus between the disease and dysregulated m6A‐mediated mRNA metabolism.

Relying on the development of big data, internet technology, and digital trade, consumers are seeking cross-border e-commerce platforms increasingly. Platform-based advertising communication, characterized by low cost and wide coverage, is notable as one of the marketing strategies favored most within general cross-border e-commerce platforms. It furnishes consumers with abundant shopping options while it exerts influence on their purchasing decisions concurrently. Using Behavioral Reasoning Theory to explore in-platform advertisements' effect on consumer purchasing decisions in cross-border e-commerce platforms, this study introduces advertising features as moderating variables. From the perspectives of self-efficacy, behavioral rationality, and inclusiveness, a behavioral reasoning model for consumer purchasing decisions is constructed. This paper focuses on cross-border e-commerce platform consumers as the research object, and uses Partial Least Squares Structural Equation Modelling (PLS-SEM) to analyze the data and validate the model. The results show that self-efficacy and behavioral rationality can have a direct effect on consumers' purchase decision-making behaviors, while self-efficacy plays a significant mediating role between behavioral reasonableness and purchase decision-making behaviors. Inclusiveness has a significant direct effect on self-efficacy and behavioral reasonableness; moreover, advertisement characteristics have a significant negative moderating effect on inclusiveness and behavioral reasonableness-adoption. This study revealed the influencing mechanisms of consumers' purchasing decision behavior on cross-border e-commerce platforms, and expands the scope of Behavioral Reasoning Theory's application. Further, it provides valuable insights for the development of cross-border e-commerce platforms that have both theoretical and practical significance.

Aiming at the problems of slow response speed, poor regulation, time-varying and nonlinear tensioning control of traditional belt conveyor tensioning system in mine, a Fuzzy-PID control (PID control based on fuzzy algorithm) method of hydraulic tensioning system of belt conveyor in mine is proposed. Firstly, a mathematical model is established according to the hydraulic tensioning device, and secondly, the hydraulic tensioning system with Fuzzy-PID controller and PID controller are simulated by the built-in Simulink simulation library of Matlab, respectively, to derive the start-up response stage of the conveyor belt tensioning force and the regulation response graph of the sudden change of tension, and to make a comparative analysis. Finally, the effectiveness of the algorithm model is verified by experimental tests. The simulation results show that the hydraulic belt conveyor tensioning Fuzzy-PID control system has better steady state performance and faster response speed not only in the start-up phase but also during the sudden change of tension. The overshoot of tension in the start-up response phase of the tensioning device is reduced by 13.5%, and the time to reach the desired value is shortened by 0.5 s. The tension change of the tensioning device is the simulated tensioning and loosening stage, when the tension is increased, the speed of regulation of the Fuzzy-PID controller is shortened by 0.4 s, and the overshoot is reduced by 4%. When the tension decreases, the regulation speed of the Fuzzy-PID controller is shortened by 0.3 s and the overshoot is reduced by 2%. The experimental results show that the effect of using Fuzzy-PID control is better and more stable with less loss. Compared with PID control, Fuzzy-PID control is more effective, the average time is reduced by 31% and the overall tendency is stable. For the mine belt conveyor such as continuous transportation equipment, Fuzzy-PID control technology for the mine belt conveyor smooth operation provides a certain guarantee, not only reduces the waste of electricity, but also reduces the maintenance cost of maintenance belt conveyor.

We used field experimental data to evaluate the ability of the agricultural production system model (APSIM) to simulate soil P availability, maize biomass and grain yield in response to P fertilizer applications on a fluvo-aquic soil in the North China Plain. Crop and soil data from a 2-year experiment with three P fertilizer application rates (0, 75 and 300 kg P2O5 ha−1) were used to calibrate the model. Sensitivity analysis was carried out to investigate the influence of APSIM SoilP parameters on the simulated P availability in soil and maize growth. Crop and soil P parameters were then derived by matching or relating the simulation results to observed crop biomass, yield, P uptake and Olsen-P in soil. The re-parameterized model was further validated against 2 years of independent data at the same sites. The re-parameterized model enabled good simulation of the maize leaf area index (LAI), biomass, grain yield, P uptake, and grain P content in response to different levels of P additions against both the calibration and validation datasets. Our results showed that APSIM needs to be re-parameterized for simulation of maize LAI dynamics through modification of leaf size curve and a reduction in the rate of leaf senescence for modern stay-green maize cultivars in China. The P concentration limits (maximum and minimum P concentrations in organs) at different stages also need to be adjusted. Our results further showed a curvilinear relationship between the measured Olsen-P concentration and simulated labile P content, which could facilitate the initialization of APSIM P pools in the NCP with Olsen-P measurements in future studies. It remains difficult to parameterize the APSIM SoilP module due to the conceptual nature of the pools and simplified conceptualization of key P transformation processes. A fundamental understanding still needs to be developed for modelling and predicting the fate of applied P fertilizers in soils with contrasting physical and chemical characteristics.

Despite its rapid growth and economic success, the sustainability of the floriculture industry as it is presently conducted is debatable, due to the huge environmental impacts it initiates and incurs. Achieving sustainability requires joint efforts from all stakeholders, a fact that is often neglected in discussions that frequently focus upon economically driven management concerns. This review attempts to raise awareness and collective responsibility among the key practitioners in floriculture by discussing its sustainability in the context of soil health, as soil is the foundation of agriculture systems. Major challenges posed to soil health arise from soil acidification and salinization stimulated by the abusive use of fertilizers. The poisoning of soil biota by pesticide residues and plastic debris due to the excessive application of pesticides and disposal of plastics is another significant issue and concern. The consequence of continuous cropping obstacles are further elucidated by the concept of plant-soil feedback. Based on these challenges, we propose the adoption and implementation of several sustainable practices including breeding stress-resistant and nutrient-efficient cultivars, making sustainable soil management a goal of floriculture production, and the recycling of plastics to overcome and mitigate the decline in soil health. The problems created by flower waste materials are highlighted and efficient treatment by biochar synthesis is suggested. We acknowledge the complexity of developing and implementing the proposed practices in floriculture as there is limited collaboration among the research and operational communities, and the policymakers. Additional research examining the impacts the floriculture industry has upon soils is needed to develop more sustainable production practices that can help resolve the current threats and to bridge the understanding gap between researchers and stakeholders in floriculture.

In this study, several 3-dimensional (3-D) parameter estimation and localization algorithms for wireless near-field (NF) sources are proposed employing the uniform circular array (UCA) structure. In the single-base-station case, the algebraic relation is demonstrated between the azimuth angle under the far-field (FF) assumption and the actual NF source firstly. Secondly, two groups of antenna pairs are selected with distances less than half the wavelength, which are called short baselines in the interferometer method. The foregoing short-baseline method is qualified to localize an NF source. In addition, a long-baseline method is also proposed with further research. Two groups of antenna pairs with distances greater than half the wavelength are selected as two long baselines. In the multiple-base-stations case, another two novel algorithms are also proposed. The first one is the centroid algorithm, which is based on the centroid calculation of three estimated source locations. And the second one is the perpendicular foot algorithm, which takes the perpendicular foot within three estimated source locations as the final positioning location. Simulation results illustrate that the proposed algorithms can achieve higher localization accuracy than the conventional 3-D Root MUSIC method. Moreover, the long-baseline method performs better than the short-baseline method. And it is also shown that the proposed perpendicular foot algorithm shows better performance than the proposed centroid algorithm.

Abstract The pore structure of coal plays a key role in controlling the storage and migration of CH4/N2. The pore structure of coal is an important indicator to measure the gas extraction capability and the gas displacement effect of N2 injection. The deformation characteristic of coal during adsorption–desorption of CH4/N2 is an important factor affecting CH4 pumpability and N2 injectability. The pore structure characteristics of low-permeability coal were obtained by fluid intrusion method and photoelectric radiation technology. The multistage and connectivity of coal pores were analyzed. Subsequently, a simultaneous test experiment of CH4/N2 adsorption–desorption and coal deformation was carried out. The deformation characteristics of coal were clarified and a coal strain model was constructed. Finally, the applicability of low-permeability coal to N2 injection for CH4 displacement technology was investigated. The results show that the micropores and transition pores of coal samples are relatively developed. The pore morphology of coal is dominated by semi-open pores. The pore structure of coal is highly complex and heterogeneous. Transition pores, mesopores and macropores of coal have good connectivity, while micropores have poor connectivity. Under constant triaxial stress, the adsorption capacity of the coal for CH4 is greater than that for N2, and the deformation capacity of the coal for CH4 adsorption is greater than that for N2 adsorption. The axial strain, circumferential strain, and volumetric strain during the entire process of CH4 and N2 adsorption/desorption in the coal can be divided into three stages. Coal adsorption–desorption deformation has the characteristics of anisotropy and gas-difference. A strain model for the adsorption–desorption of CH4/N2 from coal was established by considering the expansion stress of adsorbed gas on the coal matrix, the compression stress of free gas on the coal matrix, and the expansion stress of free gas on micropore fractures. N2 has good injectability in low-permeability coal seams and has the dual functions of improving coal seam permeability and enhancing gas flow, which can significantly improve the effectiveness of low-permeability coal seam gas control and promote the efficient utilization of gas resources.

The reservoir water level fluctuation zones (RWLFZs) and the natural riparian zones (NRZs) are two riparian ecosystems with dramatically opposite hydrological rhythms that notably influence the plant guilds. However, little is known about the discrepancies of the functional traits and niche characteristics of plant guilds in the RWLFZs and NRZs under different flooding rhythms. The aims of this study were to assess the divergent influence of natural and non-seasonal hydrological fluctuations on functional traits and niche characteristics of plant guilds. The results showed that 78 vascular plant species were identified in the riparian zones of the Xiangxi River basin. The dominant species were annuals in the two riparian ecosystems and their percentage increased temporally from 65.79% in the NRZs to 67.34% in the RWLFZs. Compared with the NRZs, the specific leaf area, vegetation coverage and the aboveground biomass in the RWLFZs significantly increased by 74%, 27% and 19.6%, respectively, while the water-use efficiency of the RWLFZ decreased by 59.6% and there was no significant difference in the net photosynthetic rate between them, which showed that annuals in the RWLFZs adopted the R adaptation strategy of being fast-growing with a short lifespan and quickly acquiring and investing resources by altering leaf morphology, including expanding the leaf area. The Simpson dominance index of RWLFZs was significantly higher than that of NRZ. Thus, counter-seasonally hydrological alterations had significant effects on functional traits of dominant species in the RWFLZs. Moreover, the highly adaptable and widely distributed species with larger niche breadths and high important values usually had a higher niche overlap value in the RWLFZs than in the NRZs, which showed that the competition in the plant communities after experiencing anti-seasonal flooding was much more intraspecific than interspecific and would facilitate the expansion of species niches. Our findings imply that the species with large niche breadths and high important values should be prioritized in ecological restoration efforts in the newly formed hydro-fluctuation zones of the TGR.

Challenging times have put organizations in a perilous and chaotic state that demands immediate resolution and calls for effective leadership to help navigate out of the crisis. In this context, we focused on psychosocial safety climate theory to investigate the influence of adaptive leadership on safety citizenship behaviors by looking at the mediating effect of readiness to change and the moderating impact of psychosocial safety climate and proactive personality, particularly in the Pakistani healthcare sector. To test the hypotheses, the data were collected from 397 employees working in the healthcare sector of Pakistan at two different times. The results of this study supported the model. The moderated path analysis revealed that psychosocial safety climate strengthens the direct effect of adaptive leadership on readiness to change, whereas the moderating impact of a proactive personality also strengthens the relationship between readiness to change and safety citizenship behaviors. Similarly, both moderators significantly moderated the indirect impact of adaptive leadership on safety citizenship behaviors via readiness to change. To conclude, the present study has significant implications for organizations and practitioners in both steady and uncertain environments.

Background: Retinoblastoma is rare but nevertheless the most common pediatric eye cancer that occurs in children under age 5. High-resolution metabolomics (HRM) is a powerful analytical approach to profile metabolic features and pathways or identify metabolite biomarkers. To date, no studies have used pre-diagnosis blood samples from retinoblastoma cases and compared them to healthy controls to elucidate early perturbations in tumor pathways. Objectives: Here, we report on metabolic profiles of neonatal blood comparing cases later in childhood diagnosed with retinoblastoma and controls. Methods: We employed untargeted metabolomics analysis using neonatal dried blood spots for 1327 children (474 retinoblastoma cases and 853 healthy controls) born in California from 1983 to 2011. Cases were selected from the California Cancer Registry and controls, frequency matched to cases by birth year, from California birth rolls. We performed high-resolution metabolomics to extract metabolic features, partial least squares discriminant analysis (PLS-DA) and logistic regression to identify features associated with disease, and Mummichog pathway analysis to characterize enriched biological pathways. Results: PLS-DA identified 1917 discriminative features associated with retinoblastoma and Mummichog identified 14 retinoblastoma-related enriched pathways including linoleate metabolism, pentose phosphate pathway, pyrimidine metabolism, fructose and mannose metabolism, vitamin A metabolism, as well as fatty acid and lipid metabolism. Interpretation: Our findings linked a retinoblastoma diagnosis in early life to newborn blood metabolome perturbations indicating alterations in inflammatory pathways and energy metabolism. Neonatal blood spots may provide a venue for early detection for this or potentially other childhood cancers.

Abstract The method of fundamental solutions with excitation source (MFS‐ES) is a reliable method for determining the eigenvalues of a two‐dimensional hollow waveguide. However, the accuracy in MFS‐ES is extremely dependent on the choice of the auxiliary boundary. In this work, to avoid the problem, a regularized MFS‐ES (RMFS‐ES) is proposed. First, to overcome the shortcomings of MFS, a regularized method of fundamental Solutions (RMFS) is proposed by borrowing the ideas of singular boundary method (SBM) and single layer regularized meshless method (SRMM). Second, the expressions for the fundamental solutions at the singularities are given by utilizing the subtraction and adding‐back technique (SAB). Third, the excitation source method is introduced to overcome the problem that the RMFS may also have spurious frequencies. Finally, to improve the accuracy of RMFS‐ES in solving polygonal waveguide eigenvalue, a wedge scheme (WS) is given. Numerical results show that the meshless RMFS‐ES is a promising numerical method for waveguide eigenvalue problems.

Background: Endoplasmic reticulum (ER) plays a crucial role in the development of organ injury caused by sepsis. Therefore, it is highly important to devise strategies that specially target ER stress for the treatment of sepsis. Previous research has shown that priming chemokines can enhance the therapeutic effects of mesenchymal stem cells (MSCs). In this study, we aimed to investigate the function and mechanism of exosomes derived from MSCs that were pretreated with IL-1β (IB-exos) in the context of septic ER stress. Methods: Mouse bone MSCs were preconditioned with or without IL-1β and the supernatant was used for exosome extraction. In vitro sepsis cell mode was induced by treating HUVECs with LPS, while in vivo sepsis model was established through cecal ligation and puncture (CLP) operation in mice. Cell viability, apoptosis, motility, and tube formation were assessed using the EDU proliferation assay, flow cytometry analysis, migration assay, and tube formation assay, respectively. The molecular mechanism was investigated using ELISA, qRT-PCR, Western blot, and immunofluorescence staining. Results: Pretreatment with IL-1β enhanced the positive impact of MSC-exos on the viability, apoptosis, motility, and tube formation ability of HUVECs. The administration of LPS or CLP increased ER stress response, but this effect was blocked by the treatment of IB-exos. Additionally, IB-exos reversed the inhibitory effects of LPS or CLP on the expression levels of SIRT1 and ERK phosphorylation. Knockdown of SIRT1 counteracted the effects of IB-exos on HUVEC cellular function and ER stress. In a mouse model, the injection of IB-exos mitigated sepsis-induced lung injury by inhibiting ER stress response through the activation of SIRT1. Conclusion: IB-exos have been found to alleviate sepsis-induced lung injury via inhibiting ER stress through the SIRT1/ERK pathway. These findings indicated that IB-exos could potentially be used as a strategy to mitigate lung injury caused by sepsis.

Abstract Cross‐modality person re‐identification is a challenging task due to the large visual appearance difference between RGB and infrared images. Existing studies mainly focus on learning local features and ignore the correlation between local features. In this paper, the Integration Graph Attention Network is proposed to learn the completed correlation between local features via the graph structure. To this end, the authors learn the coarse‐fine attention weights to aggregate the local features by considering local detail and global information. Furthermore, the Multi‐Centre Constrained Loss is proposed to optimise the feature similarity by constraining the centres of modality and identity. It simultaneously utilises three kinds of centre constraints, that is intra‐identity centre constraint, modality centre constraint, and inter‐identity centre constraint, in order to reduce the influence of modality information explicitly. The proposed method is evaluated on two standard benchmark datasets, that is SYSU‐MM01 and RegDB, and the results demonstrate that the authors’ method achieves better performance than the state‐of‐the‐art methods, for example, surpassing NFS by 4.8% and 6.0% mAP on the single‐shot setting in All‐search and Indoor‐search modes, respectively.

Abstract Background Previous studies indicated that obesity would accelerate frailty progression. However, obesity is heterogeneous by different metabolic status. The associations of metabolic heterogeneity of obesity with frailty progression remain unclear. Methods A total of 6730 participants from the China Health and Retirement Longitudinal Study (CHARLS) and 4713 from the English Longitudinal Study of Ageing (ELSA) were included at baseline. Metabolic heterogeneity of obesity was evaluated based on four obesity and metabolic phenotypes as metabolically healthy normal weight (MHNW), metabolically unhealthy normal weight (MUNW), metabolically healthy overweight/obesity (MHOO), and metabolically unhealthy overweight/obesity (MUOO). Frailty status was assessed by the frailty index (FI) ranging from 0 to 100 and frailty was defined as FI ≥ 25. Linear mixed‐effect models were used to analyse the associations of metabolic heterogeneity of obesity with frailty progression. Results In the CHARLS, MUOO and MUNW presented the accelerated FI progression with additional annual increases of 0.284 (95% CI: 0.155 to 0.413, P

Abstract Introduction Laryngeal squamous cell cancer (LSCC) is a highly malignant tumor originating from the respiratory system. Circular RNAs have been reported to be associated with the treatment and prognosis of a variety of cancers, including LSCC. Methods The expression of circBFAR, miR‐31‐5p, and collagen type V alpha 1 chain (COL5A1) in LSCC tissues and cells was detected by quantitative real‐time polymerase chain reaction. Cell counting kit 8 and 5‐Ethynyl‐2′‐deoxyuridine assays were used to detect cell proliferation. Wound healing assay and transwell assay were used to test cell migration and invasion, respectively. The protein expression in LSCC cells was detected with western blot. The relationships between miR‐31‐5p and circBFAR or COL5A1 were identified by dual‐luciferase reporter assay, RNA‐pull down assay, and immunoprecipitation assay. The effect of circBFAR on tumor growth in vivo was detected by tumor xenograft mice experiment. The protein expression of COL5A1 and KI‐67 in LSCC tissues was measured by immunohistochemistry assay. Results CircBFAR was increased in LSCC tissues and cells, and was related to advanced clinical stage and overall survival of LSCC patients. The cell viability and proliferation were inhibited by circBFAR knockdown and silencing of circBFAR blocked migration and invasion of LSCC cells. CircBFAR knockdown suppressed cell tube formation, and the protein expression of KI‐67, matrix metallopeptidase 2 (MMP2), and vascular endothelial growth factor A (VEGFA) in LSCC cells. MiR‐31‐5p was the target of circBFAR, and the inhibitory effects of circBFAR deficiency on viability, proliferation, migration, invasion, tube formation and the protein expression of KI‐67, MMP2, and VEGFA in LSCC cells were rescued by miR‐31‐5p downregulation. COL5A1 was negatively regulated by miR‐31‐5p, and was boosted in LSCC tissues and cells. COL5A1 overexpression reversed the inhibitory effects of miR‐31‐5p on LSCC cells. CircBFAR insufficiency hindered tumor growth in vivo. Conclusion CircBFAR, miR‐31‐5p, and COL5A1 in LSCC progression might provide novel therapeutic targets for LSCC clinical intervention.