Graph neural networks(GNN)have shown strong performance in node classification tasks,yet most existing models rely on uniform or shared weight aggregation,lacking flexibility in modeling the varying strength of relati...Graph neural networks(GNN)have shown strong performance in node classification tasks,yet most existing models rely on uniform or shared weight aggregation,lacking flexibility in modeling the varying strength of relationships among nodes.This paper proposes a novel graph coupling convolutional model that introduces an adaptive weighting mechanism to assign distinct importance to neighboring nodes based on their similarity to the central node.Unlike traditional methods,the proposed coupling strategy enhances the interpretability of node interactions while maintaining competitive classification performance.The model operates in the spatial domain,utilizing adjacency list structures for efficient convolution and addressing the limitations of weight sharing through a coupling-based similarity computation.Extensive experiments are conducted on five graph-structured datasets,including Cora,Citeseer,PubMed,Reddit,and BlogCatalog,as well as a custom topology dataset constructed from the Open University Learning Analytics Dataset(OULAD)educational platform.Results demonstrate that the proposed model achieves good classification accuracy,while significantly reducing training time through direct second-order neighbor fusion and data preprocessing.Moreover,analysis of neighborhood order reveals that considering third-order neighbors offers limited accuracy gains but introduces considerable computational overhead,confirming the efficiency of first-and second-order convolution in practical applications.Overall,the proposed graph coupling model offers a lightweight,interpretable,and effective framework for multi-label node classification in complex networks.展开更多
Objective:The neglect of occult lymph nodes metastasis(OLNM)is one of the pivotal causes of early non-small cell lung cancer(NSCLC)recurrence after local treatments such as stereotactic body radiotherapy(SBRT)or surge...Objective:The neglect of occult lymph nodes metastasis(OLNM)is one of the pivotal causes of early non-small cell lung cancer(NSCLC)recurrence after local treatments such as stereotactic body radiotherapy(SBRT)or surgery.This study aimed to develop and validate a computed tomography(CT)-based radiomics and deep learning(DL)fusion model for predicting non-invasive OLNM.Methods:Patients with radiologically node-negative lung adenocarcinoma from two centers were retrospectively analyzed.We developed clinical,radiomics,and radiomics-clinical models using logistic regression.A DL model was established using a three-dimensional squeeze-and-excitation residual network-34(3D SE-ResNet34)and a fusion model was created by integrating seleted clinical,radiomics features and DL features.Model performance was assessed using the area under the curve(AUC)of the receiver operating characteristic(ROC)curve,calibration curves,and decision curve analysis(DCA).Five predictive models were compared;SHapley Additive exPlanations(SHAP)and Gradient-weighted Class Activation Mapping(Grad-CAM)were employed for visualization and interpretation.Results:Overall,358 patients were included:186 in the training cohort,48 in the internal validation cohort,and 124 in the external testing cohort.The DL fusion model incorporating 3D SE-Resnet34 achieved the highest AUC of 0.947 in the training dataset,with strong performance in internal and external cohorts(AUCs of 0.903 and 0.907,respectively),outperforming single-modal DL models,clinical models,radiomics models,and radiomicsclinical combined models(DeLong test:P<0.05).DCA confirmed its clinical utility,and calibration curves demonstrated excellent agreement between predicted and observed OLNM probabilities.Features interpretation highlighted the importance of textural characteristics and the surrounding tumor regions in stratifying OLNM risk.Conclusions:The DL fusion model reliably and accurately predicts OLNM in early-stage lung adenocarcinoma,offering a non-invasive tool to refine staging and guide personalized treatment decisions.These results may aid clinicians in optimizing surgical and radiotherapy strategies.展开更多
Lymph node metastasis(LNM)is a crucial risk factor influencing an unfavorable prognosis in specific cancers.Fundamental research illuminates our understanding of tumor behavior and identifies valuable therapeutic targ...Lymph node metastasis(LNM)is a crucial risk factor influencing an unfavorable prognosis in specific cancers.Fundamental research illuminates our understanding of tumor behavior and identifies valuable therapeutic targets.Nevertheless,the exploration of fundamental theories and the validation of clinical therapies hinge on preclinical experiments.Preclinical models,in this context,serve as the conduit connecting fundamental theories to clinical outcomes.In vivo models established in animals offer a valuable platform for comprehensively observing interactions between tumor cells and organisms.Using various experimental animals,including mice,diverse methods,such as carcinogen-induced tumorigenesis,tumor cell line or human tumor transplantation,genetic engineering,and humanization,have been used effectively to construct numerous models for tumor LNM.Carcinogen-induced models simulate the entire process of tumorigenesis and metastasis.Transplantation models,using human tumor cell lines or patient-derived tumors,offer a research platform closely mirroring the histology and clinical behavior of human tumors.Genetically engineered models have been used to delve into the mechanisms of primary tumorigenesis within an intact microenvironment.Humanized models are used to overcome barriers between human and murine immune systems.Beyond mouse models,various other animal models have unique advantages and limitations,all contributing to exploring LNM.This review summarizes existing in vitro and animal preclinical models,identifies current bottlenecks in preclinical research,and offers an outlook on forthcoming preclinical models.展开更多
Objective:The current pathological diagnosis of lymph node metastasis is time-consuming,labor-intensive,and dependent on sectioning of paraffin blocks.Herein,in a prospective cohort of patients with breast cancer,we v...Objective:The current pathological diagnosis of lymph node metastasis is time-consuming,labor-intensive,and dependent on sectioning of paraffin blocks.Herein,in a prospective cohort of patients with breast cancer,we validated dynamic full-field optical coherence tomography(D-FFOCT),a virtual pathology tool integrating deep learning for nodal metastasis detection,and offering rapid and label-free histologic approximations of fresh tissues.Methods:In a prospective dual-center cohort of 155 patients with breast cancer,747 freshly bisected lymph node slides were obtained via D-FFOCT.Surgeons interpreted each slide with histopathology as the gold standard.A deep learning model was trained on 28,911 patches(corresponding to 590 slides)and tested on 7,736 patches(corresponding to 157 slides).The results were mapped to the slide level for potential intraoperative evaluation.Results:D-FFOCT strongly correlated with hematoxylin and eosin(H&E)-stained histological images.Surgeons achieved 97.10%specificity in nodal diagnosis with D-FFOCT.The performance of the artificial intelligence(AI)model was not inferior to that of human experts and had a sensitivity/specificity of 87.88%/91.94%and an area under the receiver operating characteristic curve of 0.899 at the slide level.The human–AI collaborative system reduced labor requirements by 75%and increased the specificity by 6.5%,to 98.39%.Conclusions:D-FFOCT has excellent potential as a tool for assessing lymph node metastatic status without tissue preparation or consumption.The integration of D-FFOCT with deep learning decreases labor demands and maintains high accuracy,thereby enabling streamlined nodal prediction independent of routine pathology procedures.展开更多
Rural domestic sewage treatment is critical for environmental protection.This study defines the spatial pattern of villages from the perspective of rural sewage treatment and develops an integrated decision-making sys...Rural domestic sewage treatment is critical for environmental protection.This study defines the spatial pattern of villages from the perspective of rural sewage treatment and develops an integrated decision-making system to propose a sewage treatment mode and scheme suitable for local conditions.By considering the village spatial layout and terrain factors,a decision tree model of residential density and terrain type was constructed with accuracies of 76.47%and 96.00%,respectively.Combined with binary classification probability unit regression,an appropriate sewage treatment mode for the village was determined with 87.00%accuracy.The Analytic Hierarchy Process(AHP),combined with the Technique for Order Preference(TOPSIS)by Similarity to an Ideal Solution model,formed the basis for optimal treatment process selection under different emission standards.Verification was conducted in 542 villages across three counties of the Inner Mongolia Autonomous Region,focusing on the standard effluent effect(0.3773),low investment cost(0.3196),and high standard effluent effect(0.5115)to determine the best treatment process for the same emission standard under different needs.The annual environmental and carbon emission benefits of sewage treatment in these villages were estimated.This model matches village density,geographic feature,and social development level,and provides scientific support and a theoretical basis for rural sewage treatment decision-making.展开更多
The metastatic pattern of colon cancer is typically well characterized,with initial dissemination occurring through regional lymphatics,followed by hematogenous spread.The most frequent sites of metastasis in colorect...The metastatic pattern of colon cancer is typically well characterized,with initial dissemination occurring through regional lymphatics,followed by hematogenous spread.The most frequent sites of metastasis in colorectal cancer(CRC)include regional lymph nodes(50%–70%),liver(35%–50%),lungs(21%),peritoneum(15%),and ovaries(13%).1 Isolated distant lymph node metastasis,particularly in the absence of concurrent systemic disease,is exceedingly rare in CRC.To date,only six cases of isolated axillary lymph node metastasis(ALNM)from colorectal primaries have been documented in the literature.1–6 Even more uncommon is the incidental discovery of malignant cells in anastomotic doughnuts following stoma reversal procedures.Herein,we report a rare case involving both the incidental histopathological detection of tumor cells within doughnuts during stoma closure and the subsequent development of isolated ALNM after curative resection of sigmoid colon carcinoma.展开更多
Severe combined immunodeficiency disease(SCID),characterized by profound immune system dysfunction,can lead to life-threatening infections and death.Animal models play a pivotal role in elucidating biological processe...Severe combined immunodeficiency disease(SCID),characterized by profound immune system dysfunction,can lead to life-threatening infections and death.Animal models play a pivotal role in elucidating biological processes and advancing therapeutic strategies.Recent advances in gene-editing technologies,including zincfinger nucleases(ZFNs),transcription activator-like effector nucleases(TALENs),CRISPR/Cas9,and base editing,have significantly enhanced the generation of SCID models.These models have not only deepened our understanding of disease pathophysiology but have also driven progress in cancer therapy,stem cell transplantation,organ transplantation,and infectious diseasemanagement.Thisreviewprovidesa comprehensive overview of current SCID models generated using novel gene-editing approaches,highlighting their potential applications in translational medicine and their role in advancing biomedical research.展开更多
Objectives:PSMA PET/CT(Prostate-Specific MembraneAntigen Positron Emission Tomography/Computed Tomography)offers improved accuracy in detecting lymph node invasion(LNI)in prostate cancer(PC)patients,potentially reduci...Objectives:PSMA PET/CT(Prostate-Specific MembraneAntigen Positron Emission Tomography/Computed Tomography)offers improved accuracy in detecting lymph node invasion(LNI)in prostate cancer(PC)patients,potentially reducing the need for extended pelvic lymph node dissection(ePLND).This study aims to evaluate a patient-tailored care pathway in which ePLND is performed only in patients with unfavorable intermediate-or high-risk PC who are deemed at risk for LNI based on PSMA PET/CT findings.Methods:In this interventional cohort study,81 patients were managed according to the new care pathway.ePLND was omitted in cases of negative PSMA PET/CT findings(N0M0),while those with positive PSMA PET/CT findings(N1M0)underwent ePLND.A comparator group of 81 patients was selected from a prospectively generated database for comparison.Results:The intervention group experienced a 75% reduction in the number of ePLNDs performed compared to the comparator group(p<0.001).ePLND-related complications were significantly lower in the intervention group(p=0.008).No significant difference was observed in 3-year biochemical-recurrence free survival(BRFS)between the two groups(p=0.958).Conclusion:Omitting ePLND in patients with negative PSMA PET/CT findings(N0M0)leads to a substantial reduction in the number of ePLNDs performed,resulting in a decrease in morbidity,without compromising early oncological outcomes.展开更多
We present a comprehensive description and benchmark evaluation of the global–regional chemical transport model called the Emission and Atmospheric Processes Integrated and Coupled Community(EPICC)model.The framework...We present a comprehensive description and benchmark evaluation of the global–regional chemical transport model called the Emission and Atmospheric Processes Integrated and Coupled Community(EPICC)model.The framework incorporates(1)grid configuration,(2)transport dynamics,(3)chemical mechanisms,(4)aerosol processes,(5)wet/dry deposition parameterizations,and(6)heterogeneous chemistry treatments associated with sulfate,nitrous acid(HONO)chemistry,and aerosol/cloud–photolysis interactions(APIs/CPIs).Openly shared with the atmospheric research community,the model facilitates integration of advanced physicochemical schemes to enhance simulation accuracy.Globally,the model demonstrates realistic representations of ozone(O_(3))and aerosol optical depth.The EPICC model generally demonstrates robust performance in simulating regional concentrations of O_(3) and PM_(2.5)(and its components)in China.It successfully captures vertical profiles of both global and regional O_(3).Notably,the model mitigates frequently reported sulfate underestimations in highly industrialized regions of China.The model accurately captures two regional severe pollution episodes observed in eastern China(January/June 2021).Sensitivity experiments highlight the critical roles of heterogeneous chemical mechanisms associated with sulfate,HONO chemistry,APIs,and CPIs in capturing PM_(2.5) and O_(3) concentrations in China.Improved sulfate mechanisms result in an increase of approximately 32.4%(2.8μg m^(−3))in simulated winter sulfate concentrations when observations exceed 10μg m^(−3).Enhanced HONO elevates winter O_(3) and PM_(2.5) by≤20 and≤10μg m^(−3),respectively.Overall,CPIs dominate over APIs in improving O_(3) and PM_(2.5) simulations across China.Locally,APIs mitigate PM_(2.5) and O_(3) discrepancies in the Sichuan Basin.Seasonal cloud–chemistry coupling explains the weaker impact of PM_(2.5) in summer.展开更多
Lightweight nodes are crucial for blockchain scalability,but verifying the availability of complete block data puts significant strain on bandwidth and latency.Existing data availability sampling(DAS)schemes either re...Lightweight nodes are crucial for blockchain scalability,but verifying the availability of complete block data puts significant strain on bandwidth and latency.Existing data availability sampling(DAS)schemes either require trusted setups or suffer from high communication overhead and low verification efficiency.This paper presents ISTIRDA,a DAS scheme that lets light clients certify availability by sampling small random codeword symbols.Built on ISTIR,an improved Reed–Solomon interactive oracle proof of proximity,ISTIRDA combines adaptive folding with dynamic code rate adjustment to preserve soundness while lowering communication.This paper formalizes opening consistency and prove security with bounded error in the random oracle model,giving polylogarithmic verifier queries and no trusted setup.In a prototype compared with FRIDA under equal soundness,ISTIRDA reduces communication by 40.65%to 80%.For data larger than 16 MB,ISTIRDA verifies faster and the advantage widens;at 128 MB,proofs are about 60%smaller and verification time is roughly 25%shorter,while prover overhead remains modest.In peer-to-peer emulation under injected latency and loss,ISTIRDA reaches confidence more quickly and is less sensitive to packet loss and load.These results indicate that ISTIRDA is a scalable and provably secure DAS scheme suitable for high-throughput,large-block public blockchains,substantially easing bandwidth and latency pressure on lightweight nodes.展开更多
Model evaluation using benchmark datasets is an important method to measure the capability of large language models(LLMs)in specific domains,and it is mainly used to assess the knowledge and reasoning abilities of LLM...Model evaluation using benchmark datasets is an important method to measure the capability of large language models(LLMs)in specific domains,and it is mainly used to assess the knowledge and reasoning abilities of LLMs.Therefore,in order to better assess the capability of LLMs in the agricultural domain,Agri-Eval was proposed as a benchmark for assessing the knowledge and reasoning ability of LLMs in agriculture.The assessment dataset used in Agri-Eval covered seven major disciplines in the agricultural domain:crop science,horticulture,plant protection,animal husbandry,forest science,aquaculture science,and grass science,and contained a total of 2283 questions.Among domestic general-purpose LLMs,DeepSeek R1 performed best with an accuracy rate of 75.49%.In the realm of international general-purpose LLMs,Gemini 2.0 pro exp 0205 standed out as the top performer,achieving an accuracy rate of 74.28%.As an LLMs in agriculture vertical,Shennong V2.0 outperformed all the LLMs in China,and the answer accuracy rate of agricultural knowledge exceeded that of all the existing general-purpose LLMs.The launch of Agri-Eval helped the LLM developers to comprehensively evaluate the model's capability in the field of agriculture through a variety of tasks and tests to promote the development of the LLMs in the field of agriculture.展开更多
In this paper,we establish and study a single-species logistic model with impulsive age-selective harvesting.First,we prove the ultimate boundedness of the solutions of the system.Then,we obtain conditions for the asy...In this paper,we establish and study a single-species logistic model with impulsive age-selective harvesting.First,we prove the ultimate boundedness of the solutions of the system.Then,we obtain conditions for the asymptotic stability of the trivial solution and the positive periodic solution.Finally,numerical simulations are presented to validate our results.Our results show that age-selective harvesting is more conducive to sustainable population survival than non-age-selective harvesting.展开更多
The proliferation of high-dimensional data and the widespread use of complex models present central challenges in contemporary statistics and data science.Dimension reduction and model checking,as two foundational pil...The proliferation of high-dimensional data and the widespread use of complex models present central challenges in contemporary statistics and data science.Dimension reduction and model checking,as two foundational pillars supporting scientific inference and data-driven decisionmaking,have evolved through the collective wisdom of generations of statisticians.This special issue,titled"Recent Developments in Dimension Reduction and Model Checking for regressions",not only aims to showcase cutting-edge advances in the field but also carries a distinct sense of academic homage to honor the groundbreaking and enduring contributions of Professor Lixing Zhu,a leading scholar whose work has profoundly shaped both areas.展开更多
Objective:Open retroperitoneal lymph node dissection(RPLND)is the gold-standard surgical approach for the management of metastatic testicular cancer,but robotic RPLND is becoming increasingly popular.There is limited ...Objective:Open retroperitoneal lymph node dissection(RPLND)is the gold-standard surgical approach for the management of metastatic testicular cancer,but robotic RPLND is becoming increasingly popular.There is limited research directly comparing open and robotic RPLND.The objective of this systematic review is to identify all the literature with direct comparisons between the open and robotic techniques for RPLND and to compare the two techniques.The primary outcome was peri-operative outcomes,and the secondary outcomes included oncological outcomes and patient demographics.Methods:This systematic review was prospectively registered and was conducted in accordance with the PRISMA statement.The PubMed,Embase and MEDLINE databases were searched for relevant publication from January 2006 to August 2024.Results:Eight studies,totaling 3995 patients,are included in this systematic review,with 3521 patients who underwent open RPLND and 474 who underwent robotic RPLND.For open RPLND,the mean operative duration,blood loss and length of stay were 267.8 min,475 mL and 7.3 d,respectively.For robotic RPLND,the mean operative duration,blood loss and length of stay were 334.5 min,94.6 mL and 3.7 d,respectively.Teratoma was the most common RPLND specimen pathology from both open and robotic surgeries.For open RPLND,the specimens have 13–23 nodes(26–32 mm),whereas the robotic RPLND specimens have 13–28 nodes(18–20 mm).Conclusion:This systematic review suggests that the benefitsof robotic RPLND may be associated with reduced blood loss,shorter hospitalisation and an overall lower risk of minor and major complications while maintaining oncological safety.展开更多
In their recent paper Pereira et al.(2025)claim that validation is overlooked in mapping and modelling of ecosystem services(ES).They state that“many studies lack critical evaluation of the results and no validation ...In their recent paper Pereira et al.(2025)claim that validation is overlooked in mapping and modelling of ecosystem services(ES).They state that“many studies lack critical evaluation of the results and no validation is provided”and that“the validation step is largely overlooked”.This assertion may have been true several years ago,for example,when Ochoa and Urbina-Cardona(2017)made a similar observation.However,there has been much work on ES model validation over the last decade.展开更多
This paper presents an efficient model reduction technique for linear time-varying systems based on shifted Legendre polynomials.The approach constructs approximate low-rank decomposition factors of finite-time Gramia...This paper presents an efficient model reduction technique for linear time-varying systems based on shifted Legendre polynomials.The approach constructs approximate low-rank decomposition factors of finite-time Gramians directly from the expansion coefficients of impulse responses.Leveraging these factors,we develop two model reduction algorithms that integrate the low-rank square root method with dominant subspace projection.Our method is computationally efficient and flexible,requiring only a few matrix-vector operations and a singular value decomposition of a low-dimensional matrix,thereby avoiding the need to solve differential Lyapunov equations.Numerical experiments confirm the effectiveness of the proposed approach.展开更多
Network topology obfuscation is a technique aimed at protecting critical nodes and links from disruptions such as Link Flooding Attack(LFA).Currently,there are limited topology obfuscation methods for protecting criti...Network topology obfuscation is a technique aimed at protecting critical nodes and links from disruptions such as Link Flooding Attack(LFA).Currently,there are limited topology obfuscation methods for protecting critical nodes,and the existing approaches mainly achieve obfuscation by extensivelymodifying network links,resulting in high costs.To address this issue,this paper proposes a low-cost network topology obfuscation method dedicated to critical node protection,with its core innovation lying in a lightweight obfuscation architecture based on Fake Node Clusters(FNCs).Firstly,the protected network is modeled as an undirected graph,and an adjacency matrix is constructed to quantify the network scale and structural characteristics.Then,a fake node cluster generation algorithm is designed to construct an FNC adapted to the target network.Finally,a heuristic obfuscated topology generation algorithm is proposed.By optimizing the deployment positions of Fake Nodes Clusters(FNCs)in the protected network,this algorithm effectively reduces the number of FNCs required to generate the obfuscated topology,further lowering the obfuscation cost.Extensive experiments were conducted on the public Topology Zoo dataset,categorizing network topologies by node count into small-scale([0,50)),medium-scale([50,100)),and large-scale([100,200))groups.The experimental results demonstrate that the proposed approach achieves excellent obfuscation performance,reducing the critical node recognition rate to 0%.Compared to the typical method,EigenObfu,the proposed approach also reduces obfuscation costs by an average of 97.9%,99.6%,and 99.3%for small,medium,and large-scale networks,respectively.展开更多
This study investigates the effectiveness of salicylate(SAL)as an electrolyte additive on the discharge behavior of high-purity(HP)Mg anode in an aqueous half-cell system,using an integrated approach of mathematical m...This study investigates the effectiveness of salicylate(SAL)as an electrolyte additive on the discharge behavior of high-purity(HP)Mg anode in an aqueous half-cell system,using an integrated approach of mathematical modeling and experimental analysis.A finite elementbased model is developed to elucidate the key mechanisms by which SAL influences the voltage profile and pH.Systematic electrochemical measurements,especially intermittent discharge tests combined with electrochemical impedance spectroscopy(EIS),demonstrate that SAL can enhance initial voltage stability of HP Mg anode.Moreover,the model incorporates the SAL-Mg complexation factor to describe the role of SAL in modifying the deposit film on HP Mg surface.The agreement between model predictions and experimental observations suggests that SAL facilitates the formation of compact Mg(OH)_(2) deposits and sustains a favorable pH environment within the half-cell compartment.This integrated approach provides new insights into understanding and optimizing additive effects for Mg-air batteries.展开更多
In recent years,there has been an increasing need for climate information across diverse sectors of society.This demand has arisen from the necessity to adapt to and mitigate the impacts of climate variability and cha...In recent years,there has been an increasing need for climate information across diverse sectors of society.This demand has arisen from the necessity to adapt to and mitigate the impacts of climate variability and change.Likewise,this period has seen a significant increase in our understanding of the physical processes and mechanisms that drive precipitation and its variability across different regions of Africa.By leveraging a large volume of climate model outputs,numerous studies have investigated the model representation of African precipitation as well as underlying physical processes.These studies have assessed whether the physical processes are well depicted and whether the models are fit for informing mitigation and adaptation strategies.This paper provides a review of the progress in precipitation simulation overAfrica in state-of-the-science climate models and discusses the major issues and challenges that remain.展开更多
BACKGROUND The accurate prediction of lymph node metastasis(LNM)is crucial for managing locally advanced(T3/T4)colorectal cancer(CRC).However,both traditional histopathology and standard slide-level deep learning ofte...BACKGROUND The accurate prediction of lymph node metastasis(LNM)is crucial for managing locally advanced(T3/T4)colorectal cancer(CRC).However,both traditional histopathology and standard slide-level deep learning often fail to capture the sparse and diagnostically critical features of metastatic potential.AIM To develop and validate a case-level multiple-instance learning(MIL)framework mimicking a pathologist's comprehensive review and improve T3/T4 CRC LNM prediction.METHODS The whole-slide images of 130 patients with T3/T4 CRC were retrospectively collected.A case-level MIL framework utilising the CONCH v1.5 and UNI2-h deep learning models was trained on features from all haematoxylin and eosinstained primary tumour slides for each patient.These pathological features were subsequently integrated with clinical data,and model performance was evaluated using the area under the curve(AUC).RESULTS The case-level framework demonstrated superior LNM prediction over slide-level training,with the CONCH v1.5 model achieving a mean AUC(±SD)of 0.899±0.033 vs 0.814±0.083,respectively.Integrating pathology features with clinical data further enhanced performance,yielding a top model with a mean AUC of 0.904±0.047,in sharp contrast to a clinical-only model(mean AUC 0.584±0.084).Crucially,a pathologist’s review confirmed that the model-identified high-attention regions correspond to known high-risk histopathological features.CONCLUSION A case-level MIL framework provides a superior approach for predicting LNM in advanced CRC.This method shows promise for risk stratification and therapy decisions,requiring further validation.展开更多
基金Support by Sichuan Science and Technology Program[2023YFSY0026,2023YFH0004]Guangzhou Huashang University[2024HSZD01,HS2023JYSZH01].
文摘Graph neural networks(GNN)have shown strong performance in node classification tasks,yet most existing models rely on uniform or shared weight aggregation,lacking flexibility in modeling the varying strength of relationships among nodes.This paper proposes a novel graph coupling convolutional model that introduces an adaptive weighting mechanism to assign distinct importance to neighboring nodes based on their similarity to the central node.Unlike traditional methods,the proposed coupling strategy enhances the interpretability of node interactions while maintaining competitive classification performance.The model operates in the spatial domain,utilizing adjacency list structures for efficient convolution and addressing the limitations of weight sharing through a coupling-based similarity computation.Extensive experiments are conducted on five graph-structured datasets,including Cora,Citeseer,PubMed,Reddit,and BlogCatalog,as well as a custom topology dataset constructed from the Open University Learning Analytics Dataset(OULAD)educational platform.Results demonstrate that the proposed model achieves good classification accuracy,while significantly reducing training time through direct second-order neighbor fusion and data preprocessing.Moreover,analysis of neighborhood order reveals that considering third-order neighbors offers limited accuracy gains but introduces considerable computational overhead,confirming the efficiency of first-and second-order convolution in practical applications.Overall,the proposed graph coupling model offers a lightweight,interpretable,and effective framework for multi-label node classification in complex networks.
基金supported by the National Natural Science Foundation of China(No.82272845)the Natural Science Foundation of Shandong(No.ZR2023ZD26).
文摘Objective:The neglect of occult lymph nodes metastasis(OLNM)is one of the pivotal causes of early non-small cell lung cancer(NSCLC)recurrence after local treatments such as stereotactic body radiotherapy(SBRT)or surgery.This study aimed to develop and validate a computed tomography(CT)-based radiomics and deep learning(DL)fusion model for predicting non-invasive OLNM.Methods:Patients with radiologically node-negative lung adenocarcinoma from two centers were retrospectively analyzed.We developed clinical,radiomics,and radiomics-clinical models using logistic regression.A DL model was established using a three-dimensional squeeze-and-excitation residual network-34(3D SE-ResNet34)and a fusion model was created by integrating seleted clinical,radiomics features and DL features.Model performance was assessed using the area under the curve(AUC)of the receiver operating characteristic(ROC)curve,calibration curves,and decision curve analysis(DCA).Five predictive models were compared;SHapley Additive exPlanations(SHAP)and Gradient-weighted Class Activation Mapping(Grad-CAM)were employed for visualization and interpretation.Results:Overall,358 patients were included:186 in the training cohort,48 in the internal validation cohort,and 124 in the external testing cohort.The DL fusion model incorporating 3D SE-Resnet34 achieved the highest AUC of 0.947 in the training dataset,with strong performance in internal and external cohorts(AUCs of 0.903 and 0.907,respectively),outperforming single-modal DL models,clinical models,radiomics models,and radiomicsclinical combined models(DeLong test:P<0.05).DCA confirmed its clinical utility,and calibration curves demonstrated excellent agreement between predicted and observed OLNM probabilities.Features interpretation highlighted the importance of textural characteristics and the surrounding tumor regions in stratifying OLNM risk.Conclusions:The DL fusion model reliably and accurately predicts OLNM in early-stage lung adenocarcinoma,offering a non-invasive tool to refine staging and guide personalized treatment decisions.These results may aid clinicians in optimizing surgical and radiotherapy strategies.
基金supported by the Fundamental Research Funds for the Central Universities(Wuhan University,Clinical Medicine+X,No.2042024YXB017)the Hubei Province Chinese Medicine Research Project(No.ZY2023Q015)+2 种基金the National Natural Science Foundation of China(No.61904057)the Natural Science Foundation of Hubei Province(No.2023AFB665)the Medical Young Talents Program of Hubei Province,Wuhan Young Medical Talents Training Project,China.
文摘Lymph node metastasis(LNM)is a crucial risk factor influencing an unfavorable prognosis in specific cancers.Fundamental research illuminates our understanding of tumor behavior and identifies valuable therapeutic targets.Nevertheless,the exploration of fundamental theories and the validation of clinical therapies hinge on preclinical experiments.Preclinical models,in this context,serve as the conduit connecting fundamental theories to clinical outcomes.In vivo models established in animals offer a valuable platform for comprehensively observing interactions between tumor cells and organisms.Using various experimental animals,including mice,diverse methods,such as carcinogen-induced tumorigenesis,tumor cell line or human tumor transplantation,genetic engineering,and humanization,have been used effectively to construct numerous models for tumor LNM.Carcinogen-induced models simulate the entire process of tumorigenesis and metastasis.Transplantation models,using human tumor cell lines or patient-derived tumors,offer a research platform closely mirroring the histology and clinical behavior of human tumors.Genetically engineered models have been used to delve into the mechanisms of primary tumorigenesis within an intact microenvironment.Humanized models are used to overcome barriers between human and murine immune systems.Beyond mouse models,various other animal models have unique advantages and limitations,all contributing to exploring LNM.This review summarizes existing in vitro and animal preclinical models,identifies current bottlenecks in preclinical research,and offers an outlook on forthcoming preclinical models.
基金supported by grants from the National Key Research and Development Program of China(Grant No.2024YFC3405303)Beijing Natural Science Foundation(Grant No.7242281 and 7244427)Research and Development Fund of Peking University People’s Hospital(Grant No.RDZH2024-03 and RDEB2025-25).
文摘Objective:The current pathological diagnosis of lymph node metastasis is time-consuming,labor-intensive,and dependent on sectioning of paraffin blocks.Herein,in a prospective cohort of patients with breast cancer,we validated dynamic full-field optical coherence tomography(D-FFOCT),a virtual pathology tool integrating deep learning for nodal metastasis detection,and offering rapid and label-free histologic approximations of fresh tissues.Methods:In a prospective dual-center cohort of 155 patients with breast cancer,747 freshly bisected lymph node slides were obtained via D-FFOCT.Surgeons interpreted each slide with histopathology as the gold standard.A deep learning model was trained on 28,911 patches(corresponding to 590 slides)and tested on 7,736 patches(corresponding to 157 slides).The results were mapped to the slide level for potential intraoperative evaluation.Results:D-FFOCT strongly correlated with hematoxylin and eosin(H&E)-stained histological images.Surgeons achieved 97.10%specificity in nodal diagnosis with D-FFOCT.The performance of the artificial intelligence(AI)model was not inferior to that of human experts and had a sensitivity/specificity of 87.88%/91.94%and an area under the receiver operating characteristic curve of 0.899 at the slide level.The human–AI collaborative system reduced labor requirements by 75%and increased the specificity by 6.5%,to 98.39%.Conclusions:D-FFOCT has excellent potential as a tool for assessing lymph node metastatic status without tissue preparation or consumption.The integration of D-FFOCT with deep learning decreases labor demands and maintains high accuracy,thereby enabling streamlined nodal prediction independent of routine pathology procedures.
基金supported by the Central Government Guiding Local Science and Technology Development Fund Project(No.2024SZY0343)the Joint Research Program for Ecological Conservation and High Quality Development of the Yellow River Basin(No.2022-YRUC-01-050205)+2 种基金the Higher Education Scientific Research Project of Inner Mongolia Autonomous Region(No.NJZZ23078)the project of Inner Mongolia"Prairie Talents"Engineering Innovation Entrepreneurship Talent Team,the Major Projects of Erdos Science and Technology(No.2022EEDSKJZDZX015)the Innovation Team of the Inner Mongolia Academy of Science and Technology(No.CXTD2023-01-016).
文摘Rural domestic sewage treatment is critical for environmental protection.This study defines the spatial pattern of villages from the perspective of rural sewage treatment and develops an integrated decision-making system to propose a sewage treatment mode and scheme suitable for local conditions.By considering the village spatial layout and terrain factors,a decision tree model of residential density and terrain type was constructed with accuracies of 76.47%and 96.00%,respectively.Combined with binary classification probability unit regression,an appropriate sewage treatment mode for the village was determined with 87.00%accuracy.The Analytic Hierarchy Process(AHP),combined with the Technique for Order Preference(TOPSIS)by Similarity to an Ideal Solution model,formed the basis for optimal treatment process selection under different emission standards.Verification was conducted in 542 villages across three counties of the Inner Mongolia Autonomous Region,focusing on the standard effluent effect(0.3773),low investment cost(0.3196),and high standard effluent effect(0.5115)to determine the best treatment process for the same emission standard under different needs.The annual environmental and carbon emission benefits of sewage treatment in these villages were estimated.This model matches village density,geographic feature,and social development level,and provides scientific support and a theoretical basis for rural sewage treatment decision-making.
文摘The metastatic pattern of colon cancer is typically well characterized,with initial dissemination occurring through regional lymphatics,followed by hematogenous spread.The most frequent sites of metastasis in colorectal cancer(CRC)include regional lymph nodes(50%–70%),liver(35%–50%),lungs(21%),peritoneum(15%),and ovaries(13%).1 Isolated distant lymph node metastasis,particularly in the absence of concurrent systemic disease,is exceedingly rare in CRC.To date,only six cases of isolated axillary lymph node metastasis(ALNM)from colorectal primaries have been documented in the literature.1–6 Even more uncommon is the incidental discovery of malignant cells in anastomotic doughnuts following stoma reversal procedures.Herein,we report a rare case involving both the incidental histopathological detection of tumor cells within doughnuts during stoma closure and the subsequent development of isolated ALNM after curative resection of sigmoid colon carcinoma.
基金supported by the Postdoctoral Fellowship Program of CPSF (GZC20231064)China Postdoctoral Science Foundation (2024M761345)+3 种基金Guangzhou Basic and Applied Basic Research Foundation (2024A04J6615)Scientific Research Project of Southern Medical University Stomatological Hospital (PY2023004)National Key Research and Development Program of China (2021YFA0805300)National Natural Science Foundation of China (82171244,32470564)。
文摘Severe combined immunodeficiency disease(SCID),characterized by profound immune system dysfunction,can lead to life-threatening infections and death.Animal models play a pivotal role in elucidating biological processes and advancing therapeutic strategies.Recent advances in gene-editing technologies,including zincfinger nucleases(ZFNs),transcription activator-like effector nucleases(TALENs),CRISPR/Cas9,and base editing,have significantly enhanced the generation of SCID models.These models have not only deepened our understanding of disease pathophysiology but have also driven progress in cancer therapy,stem cell transplantation,organ transplantation,and infectious diseasemanagement.Thisreviewprovidesa comprehensive overview of current SCID models generated using novel gene-editing approaches,highlighting their potential applications in translational medicine and their role in advancing biomedical research.
基金supported by a grant from Kom op tegen Kanker(Stand Up to Cancer,Belgium).
文摘Objectives:PSMA PET/CT(Prostate-Specific MembraneAntigen Positron Emission Tomography/Computed Tomography)offers improved accuracy in detecting lymph node invasion(LNI)in prostate cancer(PC)patients,potentially reducing the need for extended pelvic lymph node dissection(ePLND).This study aims to evaluate a patient-tailored care pathway in which ePLND is performed only in patients with unfavorable intermediate-or high-risk PC who are deemed at risk for LNI based on PSMA PET/CT findings.Methods:In this interventional cohort study,81 patients were managed according to the new care pathway.ePLND was omitted in cases of negative PSMA PET/CT findings(N0M0),while those with positive PSMA PET/CT findings(N1M0)underwent ePLND.A comparator group of 81 patients was selected from a prospectively generated database for comparison.Results:The intervention group experienced a 75% reduction in the number of ePLNDs performed compared to the comparator group(p<0.001).ePLND-related complications were significantly lower in the intervention group(p=0.008).No significant difference was observed in 3-year biochemical-recurrence free survival(BRFS)between the two groups(p=0.958).Conclusion:Omitting ePLND in patients with negative PSMA PET/CT findings(N0M0)leads to a substantial reduction in the number of ePLNDs performed,resulting in a decrease in morbidity,without compromising early oncological outcomes.
基金National Key Scientific and Technological Infrastructure project “Earth System Science Numerical Simulator Facility” (EarthLab)supported by the National Natural Science Foundation of China (Grant No. 92044302)the National Key Research Development Program of China (Grant No. 2022YFC3700703)
文摘We present a comprehensive description and benchmark evaluation of the global–regional chemical transport model called the Emission and Atmospheric Processes Integrated and Coupled Community(EPICC)model.The framework incorporates(1)grid configuration,(2)transport dynamics,(3)chemical mechanisms,(4)aerosol processes,(5)wet/dry deposition parameterizations,and(6)heterogeneous chemistry treatments associated with sulfate,nitrous acid(HONO)chemistry,and aerosol/cloud–photolysis interactions(APIs/CPIs).Openly shared with the atmospheric research community,the model facilitates integration of advanced physicochemical schemes to enhance simulation accuracy.Globally,the model demonstrates realistic representations of ozone(O_(3))and aerosol optical depth.The EPICC model generally demonstrates robust performance in simulating regional concentrations of O_(3) and PM_(2.5)(and its components)in China.It successfully captures vertical profiles of both global and regional O_(3).Notably,the model mitigates frequently reported sulfate underestimations in highly industrialized regions of China.The model accurately captures two regional severe pollution episodes observed in eastern China(January/June 2021).Sensitivity experiments highlight the critical roles of heterogeneous chemical mechanisms associated with sulfate,HONO chemistry,APIs,and CPIs in capturing PM_(2.5) and O_(3) concentrations in China.Improved sulfate mechanisms result in an increase of approximately 32.4%(2.8μg m^(−3))in simulated winter sulfate concentrations when observations exceed 10μg m^(−3).Enhanced HONO elevates winter O_(3) and PM_(2.5) by≤20 and≤10μg m^(−3),respectively.Overall,CPIs dominate over APIs in improving O_(3) and PM_(2.5) simulations across China.Locally,APIs mitigate PM_(2.5) and O_(3) discrepancies in the Sichuan Basin.Seasonal cloud–chemistry coupling explains the weaker impact of PM_(2.5) in summer.
基金supported in part by the Research Fund of Key Lab of Education Blockchain and Intelligent Technology,Ministry of Education(EBME25-F-08).
文摘Lightweight nodes are crucial for blockchain scalability,but verifying the availability of complete block data puts significant strain on bandwidth and latency.Existing data availability sampling(DAS)schemes either require trusted setups or suffer from high communication overhead and low verification efficiency.This paper presents ISTIRDA,a DAS scheme that lets light clients certify availability by sampling small random codeword symbols.Built on ISTIR,an improved Reed–Solomon interactive oracle proof of proximity,ISTIRDA combines adaptive folding with dynamic code rate adjustment to preserve soundness while lowering communication.This paper formalizes opening consistency and prove security with bounded error in the random oracle model,giving polylogarithmic verifier queries and no trusted setup.In a prototype compared with FRIDA under equal soundness,ISTIRDA reduces communication by 40.65%to 80%.For data larger than 16 MB,ISTIRDA verifies faster and the advantage widens;at 128 MB,proofs are about 60%smaller and verification time is roughly 25%shorter,while prover overhead remains modest.In peer-to-peer emulation under injected latency and loss,ISTIRDA reaches confidence more quickly and is less sensitive to packet loss and load.These results indicate that ISTIRDA is a scalable and provably secure DAS scheme suitable for high-throughput,large-block public blockchains,substantially easing bandwidth and latency pressure on lightweight nodes.
文摘Model evaluation using benchmark datasets is an important method to measure the capability of large language models(LLMs)in specific domains,and it is mainly used to assess the knowledge and reasoning abilities of LLMs.Therefore,in order to better assess the capability of LLMs in the agricultural domain,Agri-Eval was proposed as a benchmark for assessing the knowledge and reasoning ability of LLMs in agriculture.The assessment dataset used in Agri-Eval covered seven major disciplines in the agricultural domain:crop science,horticulture,plant protection,animal husbandry,forest science,aquaculture science,and grass science,and contained a total of 2283 questions.Among domestic general-purpose LLMs,DeepSeek R1 performed best with an accuracy rate of 75.49%.In the realm of international general-purpose LLMs,Gemini 2.0 pro exp 0205 standed out as the top performer,achieving an accuracy rate of 74.28%.As an LLMs in agriculture vertical,Shennong V2.0 outperformed all the LLMs in China,and the answer accuracy rate of agricultural knowledge exceeded that of all the existing general-purpose LLMs.The launch of Agri-Eval helped the LLM developers to comprehensively evaluate the model's capability in the field of agriculture through a variety of tasks and tests to promote the development of the LLMs in the field of agriculture.
基金Supported by the National Natural Science Foundation of China(12261018)Universities Key Laboratory of Mathematical Modeling and Data Mining in Guizhou Province(2023013)。
文摘In this paper,we establish and study a single-species logistic model with impulsive age-selective harvesting.First,we prove the ultimate boundedness of the solutions of the system.Then,we obtain conditions for the asymptotic stability of the trivial solution and the positive periodic solution.Finally,numerical simulations are presented to validate our results.Our results show that age-selective harvesting is more conducive to sustainable population survival than non-age-selective harvesting.
文摘The proliferation of high-dimensional data and the widespread use of complex models present central challenges in contemporary statistics and data science.Dimension reduction and model checking,as two foundational pillars supporting scientific inference and data-driven decisionmaking,have evolved through the collective wisdom of generations of statisticians.This special issue,titled"Recent Developments in Dimension Reduction and Model Checking for regressions",not only aims to showcase cutting-edge advances in the field but also carries a distinct sense of academic homage to honor the groundbreaking and enduring contributions of Professor Lixing Zhu,a leading scholar whose work has profoundly shaped both areas.
文摘Objective:Open retroperitoneal lymph node dissection(RPLND)is the gold-standard surgical approach for the management of metastatic testicular cancer,but robotic RPLND is becoming increasingly popular.There is limited research directly comparing open and robotic RPLND.The objective of this systematic review is to identify all the literature with direct comparisons between the open and robotic techniques for RPLND and to compare the two techniques.The primary outcome was peri-operative outcomes,and the secondary outcomes included oncological outcomes and patient demographics.Methods:This systematic review was prospectively registered and was conducted in accordance with the PRISMA statement.The PubMed,Embase and MEDLINE databases were searched for relevant publication from January 2006 to August 2024.Results:Eight studies,totaling 3995 patients,are included in this systematic review,with 3521 patients who underwent open RPLND and 474 who underwent robotic RPLND.For open RPLND,the mean operative duration,blood loss and length of stay were 267.8 min,475 mL and 7.3 d,respectively.For robotic RPLND,the mean operative duration,blood loss and length of stay were 334.5 min,94.6 mL and 3.7 d,respectively.Teratoma was the most common RPLND specimen pathology from both open and robotic surgeries.For open RPLND,the specimens have 13–23 nodes(26–32 mm),whereas the robotic RPLND specimens have 13–28 nodes(18–20 mm).Conclusion:This systematic review suggests that the benefitsof robotic RPLND may be associated with reduced blood loss,shorter hospitalisation and an overall lower risk of minor and major complications while maintaining oncological safety.
文摘In their recent paper Pereira et al.(2025)claim that validation is overlooked in mapping and modelling of ecosystem services(ES).They state that“many studies lack critical evaluation of the results and no validation is provided”and that“the validation step is largely overlooked”.This assertion may have been true several years ago,for example,when Ochoa and Urbina-Cardona(2017)made a similar observation.However,there has been much work on ES model validation over the last decade.
文摘This paper presents an efficient model reduction technique for linear time-varying systems based on shifted Legendre polynomials.The approach constructs approximate low-rank decomposition factors of finite-time Gramians directly from the expansion coefficients of impulse responses.Leveraging these factors,we develop two model reduction algorithms that integrate the low-rank square root method with dominant subspace projection.Our method is computationally efficient and flexible,requiring only a few matrix-vector operations and a singular value decomposition of a low-dimensional matrix,thereby avoiding the need to solve differential Lyapunov equations.Numerical experiments confirm the effectiveness of the proposed approach.
基金funded by the National Key Research and Development Program of China(Grant No.2022YFB3102900)the Key Science and Technology Project of Henan Province,China(No.252102211091).
文摘Network topology obfuscation is a technique aimed at protecting critical nodes and links from disruptions such as Link Flooding Attack(LFA).Currently,there are limited topology obfuscation methods for protecting critical nodes,and the existing approaches mainly achieve obfuscation by extensivelymodifying network links,resulting in high costs.To address this issue,this paper proposes a low-cost network topology obfuscation method dedicated to critical node protection,with its core innovation lying in a lightweight obfuscation architecture based on Fake Node Clusters(FNCs).Firstly,the protected network is modeled as an undirected graph,and an adjacency matrix is constructed to quantify the network scale and structural characteristics.Then,a fake node cluster generation algorithm is designed to construct an FNC adapted to the target network.Finally,a heuristic obfuscated topology generation algorithm is proposed.By optimizing the deployment positions of Fake Nodes Clusters(FNCs)in the protected network,this algorithm effectively reduces the number of FNCs required to generate the obfuscated topology,further lowering the obfuscation cost.Extensive experiments were conducted on the public Topology Zoo dataset,categorizing network topologies by node count into small-scale([0,50)),medium-scale([50,100)),and large-scale([100,200))groups.The experimental results demonstrate that the proposed approach achieves excellent obfuscation performance,reducing the critical node recognition rate to 0%.Compared to the typical method,EigenObfu,the proposed approach also reduces obfuscation costs by an average of 97.9%,99.6%,and 99.3%for small,medium,and large-scale networks,respectively.
基金the China Scholarship Council for the award of fellowship and funding No.201908510177 and No.202106050030supported by dtec.bw–Digitalization and Technology Research Center of the Bundeswehr which Dr.Deng gratefully acknowledges project DMF+1 种基金the AMABML project founded by the Zentrum für Hochleistungs-materialien(ZHM)DEZAIN project for financial support via grant from GIF,the German-Israeli Foundation for Scientific Research and Development.
文摘This study investigates the effectiveness of salicylate(SAL)as an electrolyte additive on the discharge behavior of high-purity(HP)Mg anode in an aqueous half-cell system,using an integrated approach of mathematical modeling and experimental analysis.A finite elementbased model is developed to elucidate the key mechanisms by which SAL influences the voltage profile and pH.Systematic electrochemical measurements,especially intermittent discharge tests combined with electrochemical impedance spectroscopy(EIS),demonstrate that SAL can enhance initial voltage stability of HP Mg anode.Moreover,the model incorporates the SAL-Mg complexation factor to describe the role of SAL in modifying the deposit film on HP Mg surface.The agreement between model predictions and experimental observations suggests that SAL facilitates the formation of compact Mg(OH)_(2) deposits and sustains a favorable pH environment within the half-cell compartment.This integrated approach provides new insights into understanding and optimizing additive effects for Mg-air batteries.
基金the World Climate Research Programme(WCRP),Climate Variability and Predictability(CLIVAR),and Global Energy and Water Exchanges(GEWEX)for facilitating the coordination of African monsoon researchsupport from the Center for Earth System Modeling,Analysis,and Data at the Pennsylvania State Universitythe support of the Office of Science of the U.S.Department of Energy Biological and Environmental Research as part of the Regional&Global Model Analysis(RGMA)program area。
文摘In recent years,there has been an increasing need for climate information across diverse sectors of society.This demand has arisen from the necessity to adapt to and mitigate the impacts of climate variability and change.Likewise,this period has seen a significant increase in our understanding of the physical processes and mechanisms that drive precipitation and its variability across different regions of Africa.By leveraging a large volume of climate model outputs,numerous studies have investigated the model representation of African precipitation as well as underlying physical processes.These studies have assessed whether the physical processes are well depicted and whether the models are fit for informing mitigation and adaptation strategies.This paper provides a review of the progress in precipitation simulation overAfrica in state-of-the-science climate models and discusses the major issues and challenges that remain.
基金Supported by Chongqing Medical Scientific Research Project(Joint Project of Chongqing Health Commission and Science and Technology Bureau),No.2023MSXM060.
文摘BACKGROUND The accurate prediction of lymph node metastasis(LNM)is crucial for managing locally advanced(T3/T4)colorectal cancer(CRC).However,both traditional histopathology and standard slide-level deep learning often fail to capture the sparse and diagnostically critical features of metastatic potential.AIM To develop and validate a case-level multiple-instance learning(MIL)framework mimicking a pathologist's comprehensive review and improve T3/T4 CRC LNM prediction.METHODS The whole-slide images of 130 patients with T3/T4 CRC were retrospectively collected.A case-level MIL framework utilising the CONCH v1.5 and UNI2-h deep learning models was trained on features from all haematoxylin and eosinstained primary tumour slides for each patient.These pathological features were subsequently integrated with clinical data,and model performance was evaluated using the area under the curve(AUC).RESULTS The case-level framework demonstrated superior LNM prediction over slide-level training,with the CONCH v1.5 model achieving a mean AUC(±SD)of 0.899±0.033 vs 0.814±0.083,respectively.Integrating pathology features with clinical data further enhanced performance,yielding a top model with a mean AUC of 0.904±0.047,in sharp contrast to a clinical-only model(mean AUC 0.584±0.084).Crucially,a pathologist’s review confirmed that the model-identified high-attention regions correspond to known high-risk histopathological features.CONCLUSION A case-level MIL framework provides a superior approach for predicting LNM in advanced CRC.This method shows promise for risk stratification and therapy decisions,requiring further validation.
