OBJECTIVE To form an expert consensus addressing clinical issues regarding the use of parenteral direct thrombin inhibitors （DTIs） in special populations. METHODS Led by the Sichuan Academy of Medical Sciences & Sichuan Provincial People’s Hospital（the Affiliated Hospital of UESTC）， a multidisciplinary working group was formed comprising experts from multiple fields， including clinical pharmacy， cardiac surgery， obstetrics， pediatrics and evidence-based medicine. Through literature review and the Delphi method， clinical questions regarding the efficacy and safety of parenteral DTIs used in special populations were identified. A structured design was adopted using the “Population-Intervention-Comparison-Outcome” （PICO） framework；systematic searches were conducted in CJFD， PubMed， Embase and other databases. Relevant evidence from randomized controlled trials，cohort studies and systematic reviews were included and synthesized. Evidence quality was assessed using the Grading of Recommendations Assessment，Development and Evaluation （GRADE） approach， and recommendations were formulated through three rounds of Delphi surveys and expert consensus meetings. RESULTS &CONCLUSIONS Seven clinical questions were ultimately selected （with a consensus rate exceeding 90%）， resulting in the formulation of seven recommendations on the use of parenteral DTIs in special populations， including children， pregnant women， patients with hepatic or renal impairment， patients with mesenteric venous thrombosis， and individuals with thrombophilia. These recommendations clarify the preferred agents， dosing ranges， monitoring parameters， and safety management strategies for parenteral DTIs in these special populations. This expert consensus， which is formulated based on the best available evidence， provides evidence-based guidance for standardized and individualized use of parenteral DTIs in special populations.

The initiation of adaptive immune responses relies on the precise recognition and interpretation of antigenic information. In this process, the specific binding of T cell receptors (TCRs) to peptide-major histocompatibility complex (pMHC) molecules represents one of the key molecular events in the initiation of adaptive immune responses. Accordingly, the structural features of TCR-pMHC complexes provide a fundamental basis for dissecting antigen recognition mechanisms and support rational vaccine design, therapeutic target discovery in TCR-based immunotherapy, and TCR identification and optimization. However, experimental determination of TCR-pMHC structures remains costly, time-consuming, and limited in coverage, making computational approaches essential for rapidly obtaining reliable structural information. Computational methods for predicting the structures of TCR-pMHC complexes have advanced rapidly in recent years, driven by progress in deep learning-based modeling frameworks and the increasing availability of structural and sequence resources. Despite these developments, most existing tools do not adequately distinguish the key structural and biophysical differences between MHC class I (MHC-I) and MHC class II (MHC-II) complexes during model construction. As a consequence, their predictive performance differs substantially between class I and class II complexes. In general, structural predictions for class I complexes outperform those for class II complexes. This discrepancy may be related to several fundamental differences between the two systems, including the architecture of the peptide-binding groove, the distribution of peptide lengths, and the properties of peptide flanking residues (PFRs). Compared with MHC-I molecules, MHC-II molecules usually bind longer antigenic peptides, which typically range from 13 to 25 amino acids in length. PFRs at both termini of these peptides participate in regulating the overall conformation of TCR-pMHC class II complexes and exert a pronounced effect on the geometric and physicochemical characteristics of the TCR-pMHC binding interface. Furthermore, within the TCR recognition interface, the complementarity-determining regions (CDRs) consist of segments that differ markedly in conformational behavior. They commonly include regions that are relatively rigid and structurally stable, together with highly flexible segments exhibiting substantial conformational plasticity. These rigidity-flexibility features constitute an essential structural basis enabling TCRs to recognize diverse peptide-MHC ligands and to accommodate conformational heterogeneity at the interface. However, many current modeling tools, in an effort to enforce global conformational stability or reduce structural noise, tend to over-constrain intrinsically flexible regions. Such oversimplification may lead to inappropriate rigidification of flexible CDR loops, resulting in local structural distortions, compromised interface geometry, or even complete modeling failure for specific complexes. Against this background, the review approaches the field from the perspective of computational differences between MHC-I and MHC-II complexes. We first systematically organize and summarize available resources related to TCRs and pMHCs, including structural datasets, sequence databases, prediction tools, and benchmarking studies. We then focus on five representative tools capable of predicting both class I and class II complexes—AlphaFold2, AlphaFold3, TCRmodel2, tFold-TCR, and TCR-pHLA_ModellerS. After excluding structures present in the training sets of these tools, we constructed a benchmark dataset comprising 25 class I and 10 class II TCR-pMHC complexes in the bound state and conducted a systematic evaluation using this dataset. We first employ widely used general evaluation metrics, including All-Atom Root Mean Square Deviation (All-Atom RMSD), Backbone RMSD, Template Modeling score (TM-score), and DockQ, to assess the global conformational accuracy and interface modeling quality of class I and class II complexes. For class II complexes, we propose for the first time a peptide flanking residue deviation index, including the PFRs-Deviation Index (PFRs-DI), N-PFR-Deviation Index (N-PFR-DI), and C-PFR-Deviation Index (C-PFR-DI), to quantitatively characterize conformational deviations in PFRs. In addition, we propose the CDR conformational consistency index (CCC) designed to qualitatively evaluate the ability of prediction tools to capture TCR CDR conformational flexibility. These metrics collectively assess a tool’s ability to model both overall conformation and critical functional regions, thereby addressing the limitations of existing evaluation criteria that overemphasize global structure while inadequately capturing modeling quality in key functional areas. This establishes a unified analytical framework for MHC-I and MHC-II complexes to guide data resource selection, modeling strategy formulation, and evaluation system development. The framework further advances computational modeling and provides crucial support for multi-scale analysis of TCR-pMHC recognition mechanisms and their biological functions.

The initiation of adaptive immune responses relies on the precise recognition and interpretation of antigenic information. In this process, the specific binding of T cell receptors (TCRs) to peptide-major histocompatibility complex (pMHC) molecules represents one of the key molecular events in the initiation of adaptive immune responses. Accordingly, the structural features of TCR-pMHC complexes provide a fundamental basis for dissecting antigen recognition mechanisms and support rational vaccine design, therapeutic target discovery in TCR-based immunotherapy, and TCR identification and optimization. However, experimental determination of TCR-pMHC structures remains costly, time-consuming, and limited in coverage, making computational approaches essential for rapidly obtaining reliable structural information. Computational methods for predicting the structures of TCR-pMHC complexes have advanced rapidly in recent years, driven by progress in deep learning-based modeling frameworks and the increasing availability of structural and sequence resources. Despite these developments, most existing tools do not adequately distinguish the key structural and biophysical differences between MHC class I (MHC-I) and MHC class II (MHC-II) complexes during model construction. As a consequence, their predictive performance differs substantially between class I and class II complexes. In general, structural predictions for class I complexes outperform those for class II complexes. This discrepancy may be related to several fundamental differences between the two systems, including the architecture of the peptide-binding groove, the distribution of peptide lengths, and the properties of peptide flanking residues (PFRs). Compared with MHC-I molecules, MHC-II molecules usually bind longer antigenic peptides, which typically range from 13 to 25 amino acids in length. PFRs at both termini of these peptides participate in regulating the overall conformation of TCR-pMHC class II complexes and exert a pronounced effect on the geometric and physicochemical characteristics of the TCR-pMHC binding interface. Furthermore, within the TCR recognition interface, the complementarity-determining regions (CDRs) consist of segments that differ markedly in conformational behavior. They commonly include regions that are relatively rigid and structurally stable, together with highly flexible segments exhibiting substantial conformational plasticity. These rigidity-flexibility features constitute an essential structural basis enabling TCRs to recognize diverse peptide-MHC ligands and to accommodate conformational heterogeneity at the interface. However, many current modeling tools, in an effort to enforce global conformational stability or reduce structural noise, tend to over-constrain intrinsically flexible regions. Such oversimplification may lead to inappropriate rigidification of flexible CDR loops, resulting in local structural distortions, compromised interface geometry, or even complete modeling failure for specific complexes. Against this background, the review approaches the field from the perspective of computational differences between MHC-I and MHC-II complexes. We first systematically organize and summarize available resources related to TCRs and pMHCs, including structural datasets, sequence databases, prediction tools, and benchmarking studies. We then focus on five representative tools capable of predicting both class I and class II complexes—AlphaFold2, AlphaFold3, TCRmodel2, tFold-TCR, and TCR-pHLA_ModellerS. After excluding structures present in the training sets of these tools, we constructed a benchmark dataset comprising 25 class I and 10 class II TCR-pMHC complexes in the bound state and conducted a systematic evaluation using this dataset. We first employ widely used general evaluation metrics, including All-Atom Root Mean Square Deviation (All-Atom RMSD), Backbone RMSD, Template Modeling score (TM-score), and DockQ, to assess the global conformational accuracy and interface modeling quality of class I and class II complexes. For class II complexes, we propose for the first time a peptide flanking residue deviation index, including the PFRs-Deviation Index (PFRs-DI), N-PFR-Deviation Index (N-PFR-DI), and C-PFR-Deviation Index (C-PFR-DI), to quantitatively characterize conformational deviations in PFRs. In addition, we propose the CDR conformational consistency index (CCC) designed to qualitatively evaluate the ability of prediction tools to capture TCR CDR conformational flexibility. These metrics collectively assess a tool’s ability to model both overall conformation and critical functional regions, thereby addressing the limitations of existing evaluation criteria that overemphasize global structure while inadequately capturing modeling quality in key functional areas. This establishes a unified analytical framework for MHC-I and MHC-II complexes to guide data resource selection, modeling strategy formulation, and evaluation system development. The framework further advances computational modeling and provides crucial support for multi-scale analysis of TCR-pMHC recognition mechanisms and their biological functions.

ObjectiveThis paper aims to observe the syndrome improvement and negative antigen conversion rate of Qingxuan Daozhi formula in the treatment of influenza in children （heat accumulation in the lung and stomach syndrome）. MethodsThrough a multi-center randomized controlled methodology design，confirmed influenza cases were collected from October 2022 to April 2023 in the pediatrics department of eight hospitals，such as Dongfang Hospital of Beijing University of Chinese Medicine. A total of 180 children with influenza and heat accumulation in the lung and stomach syndrome conforming to the standard were recruited through the clinic. The sick children meeting the inclusion criteria were randomly divided into groups by a block-randomized method. The children in the experimental group were treated with Qingxuan Daozhi formula for five days，and those in the control group were treated with Oseltamivir Phosphate Granules for five days. The primary efficacy indicator was the negative conversion rate of influenza antigen detection. Secondary efficacy indicators were the Canadian acute respiratory illness and flu scale （CARIFS） and the incidence of complications，severe cases， and critical cases. Follow-up observation was conducted on the day of enrollment，48 hours after medication，72 hours after medication， and （6+1） d after medication. ResultsOne hundred and eighty participants were randomly assigned to the experimental group （90 cases） or the control group （90 cases）. All participants were followed up during the study. Comparison of influenza antigen detection results in the primary efficacy indicators showed that the average time of negative influenza antigen conversion in the experimental group was （5.29±1.25） d，and that in the control group was （5.40±1.68） d，without a statistically significant difference. After five days of intervention，52 cases in the experimental group and 51 cases in the control group converted to negative，without a statistically significant difference. CARIFS score results in the secondary efficacy indicators showed that during 72 hours after intervention，there were statistically significant differences between the experimental group and the control group in three dimensions， including headache，muscle soreness， and the need for extra care （P<0.05）. On the （6+1） days after the intervention，the differences in both the experimental group and the control group were statistically significant in 10 dimensions， including sore throat，bad sleep，uncomfortable feeling，poor spirit and fatigue，crying more than usual，the need for extra care，symptom，function，influence on parents，and total score （P<0.05）. The comparison results within the group in the dimensional scores of symptom， function， and influence on parents，as well as the CARIFS total score showed that with the delay of follow-up time，scores of both groups decreased significantly，with a statistically significant difference （P<0.01）. Inter-group comparison results showed that the mean score of the experimental group was higher than that of the control group at the time of enrollment. With the progress of intervention，the score of the experimental group was significantly decreased compared with that of the control group. At the end of follow-up，the mean score of the experimental group was lower than that of the control group，with no statistically significant difference. In terms of the incidence of complications，severe cases， and critical cases， there were no complications，severe cases， and critical cases in the two groups，without a statistically significant difference. ConclusionThe symptom improvement effect and negative antigen conversion rate of Qingxuan Daozhi formula in the treatment of influenza in children （heat accumulation in the lung and stomach syndrome） are not inferior to Oseltamivir Phosphate granules， and children's acceptance is better. It can be more widely used in clinical treatment of influenza in children （heat accumulation in the lung and stomach syndrome）.

Multidrug-resistant (MDR) bacterial infections have emerged as a serious challenge of global public health crisis. The overuse and misuse of conventional antibiotics have dramatically accelerated the emergence, evolution and worldwide spread of drug-resistant bacterial strains, necessitating urgent exploration of novel antibacterial strategies. Bacteriophages serve as natural bacterial predators offering distinct advantages including high host specificity, autonomous self-replication capabilities and cost-effective large-scale production. However, wild-type phages present significant clinical limitations due to their narrow host ranges, susceptibility to rapid immune clearance and poor penetration of bacterial biofilms, which severely restrict their therapeutic applications. The convergence of synthetic biology, nanotechnology and advanced gene editing technologies has accelerated the development of engineered bacteriophage platforms, providing programmable, scalable and clinically translatable pathways to overcome these inherent biological constraints. Here, we systematically delineate four fundamental strategies for engineered bacteriophage development. Chemical modification utilizes reactive functional groups such as amino, carboxyl and thiol moieties on capsid proteins through esterification, amidation or click chemistry reactions to achieve precise drug conjugation and surface functionalization. In vivo editing encompasses ultraviolet or chemical mutagenesis for random mutation induction, homologous recombination for targeted genetic alterations, recombineering methodologies including electroporation-mediated bacteriophage recombination engineering, and CRISPR-Cas systems for precise genome editing to enable exact genetic reconstruction and host range reprogramming. In vitro synthesis leverages genome engineering platforms where intact phage genomes are transferred into yeast or host bacteria to facilitate highly efficient homologous recombination, enabling large DNA fragment assembly and cross-gene host range expansion without bacterial toxicity constraints. Directed evolution combines artificial selection through mutation library screening with rational design approaches involving chimeric receptor binding protein construction or site-specific mutagenesis, effectively balancing the discovery of unknown adaptive pathways with targeted host specificity modification. Moreover, we comprehensively discuss therapeutic applications across diverse clinical scenarios. Engineered bacteriophage effectively disrupt bacterial biofilms through sophisticated functionalized delivery platforms including nanozyme-conjugated phages, phage-liposome nanoconjugates and bio-responsive hydrogels, demonstrating significantly enhanced bactericidal efficiency compared to unmodified free phages. These bioengineered vectors attenuate bacterial virulence and resensitize pathogens to antibiotics by delivering CRISPR-Cas systems or base editors to disrupt critical virulence factors such as pili, capsule synthesis machineries and quorum sensing systems, or by inactivating antibiotic resistance determinants including beta-lactamase genes. As an intelligent nanomedicine delivery platform, engineered bacteriophage enable precise pathogen elimination an through photocatalytic reactive oxygen species generation, immunomodulatory interventions, or controlled release of antibacterial drugs. Furthermore, oral administration of engineered bacteriophage facilitates microbiota modulation, which selectively eliminate intestinal pathogens while preserve beneficial commensal microbiota, thereby restoring microbial community balance and preventing complications associated with dysbiosis. Finally, we critically analyze persistent challenges including host strain matching complexity, evolution of bacterial resistance mechanisms, pharmacokinetic optimization requirements, optimal administration route selection, large-scale production quality control standards and clinical dosing determination protocols. Through multidisciplinary integration of synthetic biology, infectious disease medicine and immunology, future translational medicine studies of bacteriophage should establish comprehensive technical platforms encompassing rapid phage screening, intelligent rational design, rigorous in vivo evaluation and standardized clinical validation processes, ultimately advancing engineered bacteriophage from laboratory innovations to clinically approved therapeutics for effectively combating MDR bacterial infections.

Objective To investigate the epidemiological characteristics of malaria and provide a basis for developing improved prevention and control measures. Methods Data were obtained from the Chinese Disease Prevention and Control Information System. Malaria surveillance data for Nanning City from January 1,2023, to December 31,2024, were exported from the Infectious Disease Reporting Information Management Subsystem. The characteristics of the two groups of malaria cases were compared. Results A total of 103 imported malaria cases were reported in Nanning City in 2024, representing a 38.32% decrease compared with the same period of the previous year. No statistically significant difference were observed between cases reported in 2023 and 2024 in terms of average age, gender ratio, proportion of parasite species, and monthly reporting distribution;however, statistically significant differences were found in the proportion of reporting areas and current residence areas(χ2= 13.572 and 10.355, respectively; P = 0.001 and 0.035, respectively). The proportion of cases reported in Shanglin County and the proportion of cases residing in Shanglin County were both lower than those during the same period of the previous year. Conclusions The high aggregation of imported malaria cases in Nanning City has decreased. Medical institutions in areas other than Shanglin County should strengthen their vigilance against malaria.

ObjectiveTo explore the mechanisms through which Si Junzitang （SJZD） ameliorates Alzheimer's disease （AD） induced by scopolamine （SCOP） in mice and the PC12 cell model induced by H₂O₂ based on the Kelch-like ECH-associated protein 1 （Keap1）/nuclear factor erythroid 2-related factor 2 （Nrf2）/heme oxygenase-1 （HO-1） signaling pathway. MethodsIn the animal experiments， an AD model was established in mice by intraperitoneal injection of SCOP （3 mg·kg^-1）. Morris water maze and open field tests （OFT） were conducted to assess learning and memory abilities. Hematoxylin-eosin （HE） staining and Nissl staining were performed to observe pathological changes in neurons. Immunofluorescence was used to detect amyloid β-protein_1-42 （Aβ_1-42） expression， and immunohistochemistry was employed to detect phosphorylated （p）-Tau expression. Transmission electron microscopy （TEM） was employed to observe ultrastructural changes in hippocampal neurons and synapses. Biochemical methods were used to measure the levels of acetylcholine （ACh）， acetylcholinesterase （AChE）， superoxide dismutase （SOD）， malondialdehyde （MDA）， catalase （CAT）， and lactate dehydrogenase （LDH）. Real-time PCR and Western blot were employed to measure the mRNA and protein levels of molecules in the Keap1/Nrf2/HO-1 pathway in the hippocampus. In the cell experiments， a PC12 cell model of oxidative damage model was established with H₂O₂. Cell count kit-8 （CCK-8） assays and flow cytometry were adopted to measure cell viability and apoptosis rates， and Western blot was employed to quantify the expression levels of proteins in the Keap1/Nrf2/HO-1 pathway. ResultsThe animal experiments showed that compared with the model group， the SJZD and donepezil groups showed shortened escape latency （P<0.01）， increased time in the target quadrant and platform crossings， and increased movement distance and duration in the central area of the open field （P<0.05， P<0.01）. HE and Nissl staining showed more organized neurons and increased Nissl bodies in the drug intervention groups （P<0.05， P<0.01）， and the Aβ_1-42 and p-Tau expression levels were downregulated （P<0.05， P<0.01）. TEM revealed reduced ultrastructural damage in hippocampal neurons and synapses in the drug intervention groups. In addition， the drug intervention groups showed declined levels of MDA， LDH， and AChE （P<0.05， P<0.01）， elevated levels of SOD， CAT， and ACh （P<0.05， P<0.01）， reduced Keap1 expression and increased Nrf2， HO-1， and NQO1 expression in the hippocampus （P<0.05， P<0.01）. The cell experiments showed that compared with the model group， the SJZD-containing serum increased the cell viability （P<0.05， P<0.01）， and decreased total apoptosis rates （P<0.05， P<0.01）. The drug intervention groups showed upregulated protein levels of Nrf2 and HO-1 and downregulated protein level of Keap1 （P<0.05， P<0.01）. ConclusionSJZD demonstrates protective effects against SCOP-induced AD in mice and H₂O₂-induced damage in PC12 cells through antioxidant mechanisms mediated by the Keap1/Nrf2/HO-1 pathway.

ObjectiveTo construct a risk prediction model for carotid artery plaques in high-risk populations of stroke based on five machine learning methods. MethodsThe clinical information of the high-risk population of stroke was collected. Factor analysis and statistical analysis of syndrome elements were conducted on their traditional Chinese medicine （TCM） symptoms and tongue and pulse manifestations. On the basis of the results of factor analysis and variable screening， five machine learning methods-classification and regression tree （CART） decision tree， support vector machine （SVM）， back propagation（BP） neural network， logistic regression， and random forest-were used to construct the risk prediction model for carotid artery plaques. ResultsThe most common TCM syndrome elements in the high-risk population of stroke was Qi deficiency. The scores of Qi deficiency， Yin deficiency， and Yang deficiency in the population with carotid artery plaques were higher than those without carotid artery plaques （P<0.05）. The CART decision tree， SVM， logistic regression， BP neural network， and random forest models showed the areas under the receiver operating characteristic （ROC） curves of 0.71， 0.75， 0.76， 0.76， and 0.75， the accuracy rates of 68.94%， 69.27%， 69.44%， 70.10%， and 69.60%， the precision rates of 68.56%， 68.53%， 68.75%， 69.74%， and 69.42%， the recall rates of 68.91%， 67.93%， 67.92%， 68.10%， and 67.31%， and the F₁ values of 0.69， 0.68， 0.68， 0.68， and 0.68， respectively. ConclusionAmong the high-risk population of stroke， the most frequently distributed TCM syndrome element is Qi deficiency， with the rest mainly being fire heat， Yin deficiency， Yang deficiency， phlegm dampness， blood stasis， and Qi stagnation. Deficiency syndrome may be a major factor leading to carotid artery plaques in the high-risk population of stroke. The BP neural network model demonstrates better performance in predicting the risk of carotid artery plaques in the high-risk population of stroke. People with carotid artery plaques are more likely to present with symptoms such as a heavy head， dizziness， headache， and thready pulse. The primary community benefits more widely when the BP neural network model is adopted to predict the risk of carotid artery plaques in the high-risk population of stroke over 40 years old.

ObjectiveTo construct a risk prediction model for carotid artery plaques in high-risk populations of stroke based on five machine learning methods. MethodsThe clinical information of the high-risk population of stroke was collected. Factor analysis and statistical analysis of syndrome elements were conducted on their traditional Chinese medicine （TCM） symptoms and tongue and pulse manifestations. On the basis of the results of factor analysis and variable screening， five machine learning methods-classification and regression tree （CART） decision tree， support vector machine （SVM）， back propagation（BP） neural network， logistic regression， and random forest-were used to construct the risk prediction model for carotid artery plaques. ResultsThe most common TCM syndrome elements in the high-risk population of stroke was Qi deficiency. The scores of Qi deficiency， Yin deficiency， and Yang deficiency in the population with carotid artery plaques were higher than those without carotid artery plaques （P<0.05）. The CART decision tree， SVM， logistic regression， BP neural network， and random forest models showed the areas under the receiver operating characteristic （ROC） curves of 0.71， 0.75， 0.76， 0.76， and 0.75， the accuracy rates of 68.94%， 69.27%， 69.44%， 70.10%， and 69.60%， the precision rates of 68.56%， 68.53%， 68.75%， 69.74%， and 69.42%， the recall rates of 68.91%， 67.93%， 67.92%， 68.10%， and 67.31%， and the F₁ values of 0.69， 0.68， 0.68， 0.68， and 0.68， respectively. ConclusionAmong the high-risk population of stroke， the most frequently distributed TCM syndrome element is Qi deficiency， with the rest mainly being fire heat， Yin deficiency， Yang deficiency， phlegm dampness， blood stasis， and Qi stagnation. Deficiency syndrome may be a major factor leading to carotid artery plaques in the high-risk population of stroke. The BP neural network model demonstrates better performance in predicting the risk of carotid artery plaques in the high-risk population of stroke. People with carotid artery plaques are more likely to present with symptoms such as a heavy head， dizziness， headache， and thready pulse. The primary community benefits more widely when the BP neural network model is adopted to predict the risk of carotid artery plaques in the high-risk population of stroke over 40 years old.

Objective To construct and verify an early prediction model for visual benefit of diabetic macular edema(DME)after anti-vascular endothelial growth factor(VEGF)treatment based on clinical data,optical coherence tomo-graphy angiography(OCTA),serum brain tissue aquaporin-4(AQP4)mRNA and total bilirubin(TBIL)levels.Methods A total of 480 patients(480 eyes)with DME treated in the First People's Hospital of Zigong City from October 2021 to March 2024 were selected and divided into a modeling set(320 cases)and a validation set(160 cases)at a ratio of 2∶1.According to the visual benefit after anti-VEGF treatment,patients in the modeling set were further divided into a benefit group(80 cases)and a non-benefit group(240 cases).The baseline data of the two groups of patients were collected,and the factors influencing visual benefits in DME patients after anti-VEGF treatment were analyzed.An early prediction model was constructed and validated both internally and externally.Results The inter-group comparison results showed that the diabetes duration in the non-benefit group was longer than that in the benefit group(P＜0.05).The proportion of smokers,the best corrected visual acuity(BCVA),the minimum resolution angle(logMAR)vision,hemoglobin A1c(HbAlc)and AQP4 mRNA levels were higher in the non-benefit group than those in the benefit group(all P＜0.05).The foveal retinal deep capillary plexus blood flow density(DCP-VD),central macular thickness(CMT),and TBIL levels were lower in the non-benefit group than those in the benefit group(all P＜0.05).The least absolute shrinkage and selection operator(LAS-SO)-Logistic regression analysis showed that the factors influencing visual benefit in DME patients after anti-VEGF treat-ment were CMT,BCVA(logMAR),HbAlc,AQP4 mRNA levels,foveal DCP-VD,and TBIL levels.The predictive risk con-sistency index of the nomogram model constructed based on the above-mentioned influencing factors for visual benefit pre-diction after anti-VEGF treatment was 0.844.The receiver operating characteristic(ROC)curve showed that the area un-der the ROC curve(AUC)of the model was 0.844(95％ CI:0.797-0.891)in the modeling set and 0.898(95％ CI:0.847-0.949)in the validation set.The decision analysis curve showed that when the high-risk threshold of the modeling set ranged between 0 and 82％ and that of the validation set ranged between 0 and 100％,the model could bring net clinical benefits.Conclusion CMT,BCVA(logMAR),HbAlc,AQP4 mRNA levels,foveal DCP-VD,and TBIL levels are the fac-tors influencing visual benefit in DME patients after anti-VEGF treatment.The visual benefit prediction model constructed based on these factors has high accuracy and stability,and can be used as an effective tool for clinical prediction of visual benefit after treatment.