Zhangsheng Yu
Professor Adjunct of BiostatisticsCards
About
Research
Publications
2025
A multimodal AI model for precision prognosis in clear cell renal cell carcinoma: A multicenter study
Zang X, Xia Y, Xiao H, Luo H, Si M, Hou N, Haoni A, Chen T, Liu Z, Pu X, Zi X, Xu L, Zhu J, Xu Z, Wang J, Wang Z, Xia J, Cao D, Yin Y, Wang J, Wu X, Kong W, Huang J, Zhang J, Chen Y, Huang Y, Leung D, Teoh J, Wang K, Liang C, Zheng J, Yu Z, Zhai W. A multimodal AI model for precision prognosis in clear cell renal cell carcinoma: A multicenter study. Npj Digital Medicine 2025, 8: 668. PMID: 41249481, PMCID: PMC12623892, DOI: 10.1038/s41746-025-02034-x.Peer-Reviewed Original ResearchCell renal cell carcinomaRenal cell carcinomaAdjuvant therapyCell carcinomaInadequate adjuvant therapyRisk of recurrenceNon-recurrent patientsClinical toolRecurrence risk stratificationC-index valuesTherapeutic decision-makingRecurrence scoreRecurrent patientsClinicopathological factorsMulticenter studyClinical featuresRisk stratificationPrecision prognosisC-indexMolecular profilingClinical feasibilityLow riskHigh riskPrognostic modelPatientsDeep partially linear transformation model for right-censored survival data
Yin J, Zhang Y, Yu Z. Deep partially linear transformation model for right-censored survival data. Biometrics 2025, 81: ujaf126. PMID: 41165358, DOI: 10.1093/biomtc/ujaf126.Peer-Reviewed Original ResearchConceptsLinear transformation modelRight-censored survival dataSemiparametric transformation modelsRight-censored dataAnalysis of survival dataMaximum likelihood estimationOverall convergence rateSurvival dataComprehensive simulation studySemiparametric efficiencyAsymptotic normalityLikelihood estimationParametric estimationCurse of dimensionalityPH assumptionConvergence rateEstimation procedureSimulation studyTransformation modelRegression frameworkDeep neural network estimatorSurvival modelsEstimationMinimaxNetwork estimationClinical Features of Chinese Patients with Thyroid Eye Disease: A Multicenter Retrospective Study
Lei C, Lyu X, Ren Y, Wei D, Zhang S, Zhang Y, Wang L, Liu L, Wen J, Liu X, Lin C, Lu W, Liu J, Li R, Zhang S, Song X, Yu Z, Bahn R, Zhou H. Clinical Features of Chinese Patients with Thyroid Eye Disease: A Multicenter Retrospective Study. Thyroid 2025, 35: 1187-1197. PMID: 40658134, DOI: 10.1177/10507256251359559.Peer-Reviewed Original ResearchThyroid eye diseaseSight-threatening thyroid eye diseaseTED severityRetrospective multicenter studyClinical featuresTreatment responseIntravenous glucocorticoidsMulticenter studyEye diseaseClinical features of Chinese patientsChinese cohortSevere thyroid eye diseaseFeatures of Chinese patientsClinical activity scoreMulticenter retrospective studyDegree of exophthalmosDebilitating autoimmune disorderModerate-to-severeConclusions:Area under the receiver operating curveReceiver operating curveDiplopia scoreSight-threateningOphthalmology departmentThyroid dysfunctionAI‐Driven De Novo Design of Ultra Long‐Acting GLP‐1 Receptor Agonists
Wei T, Ma J, Cui X, Lin J, Zheng Z, Cheng L, Cui T, Lin X, Zhu J, Ran X, Hong X, Johnston L, Yu Z, Chen H. AI‐Driven De Novo Design of Ultra Long‐Acting GLP‐1 Receptor Agonists. Advanced Science 2025, 12: e07044. PMID: 40787887, PMCID: PMC12561408, DOI: 10.1002/advs.202507044.Peer-Reviewed Original ResearchFunctional screeningDe novo designProtein designObese mouse modelImportant peptidesMouse modelIn vitro validationPeptide drug designPeptide candidatesLong-acting GLP-1 receptor agonistDiabetic mouse modelGLP-1 receptor agonistsPeptideWeight loss efficacyDrug designIn vivo experimentsNovel treatment avenuesLower blood glucose levelsBlood glucose levelsReceptor agonistsGLP-1RAGLP-1RAsEnhancing gastric cancer prognosis prediction via multi-step multi-modality ensemble survival modeling of HE-stained images and mIHC data
Gao R, Yuan X, Sun Y, Wang Y, Xia Y, Wei T, Xu D, Yu Z. Enhancing gastric cancer prognosis prediction via multi-step multi-modality ensemble survival modeling of HE-stained images and mIHC data. Computer Methods And Programs In Biomedicine 2025, 270: 108971. PMID: 40714417, DOI: 10.1016/j.cmpb.2025.108971.Peer-Reviewed Original ResearchArtificial intelligence in prostate cancer
Li W, Hu R, Zhang Q, Yu Z, Deng L, Zhu X, Xia Y, Song Z, Cimadamore A, Chen F, Lopez-Beltran A, Montironi R, Cheng L, Chen R. Artificial intelligence in prostate cancer. Chinese Medical Journal 2025, 138: 1769-1782. PMID: 40629505, PMCID: PMC12321470, DOI: 10.1097/cm9.0000000000003689.Peer-Reviewed Original Researchti-scMR: trajectory-inference-based dynamic single-cell Mendelian randomization identifies causal genes underlying phenotypic differences
Sun J, Dong Q, Wei J, Gao Y, Yu Z, Hu X, Zhang Y. ti-scMR: trajectory-inference-based dynamic single-cell Mendelian randomization identifies causal genes underlying phenotypic differences. NAR Genomics And Bioinformatics 2025, 7: lqaf082. PMID: 40630931, PMCID: PMC12231591, DOI: 10.1093/nargab/lqaf082.Peer-Reviewed Original ResearchConceptsMendelian randomizationCausal genesSingle-cell expressionGenetic instrumental variablesPhenotypic differencesIndividual phenotypesPotential causal genesPresence of confoundersEffects of gene expressionGenotype to phenotypeCausal pathwaysSingle-cell datasetsDifferential expression analysisExpression of genesPopulation genomicsSingle-cell transcriptomicsComplex traitsPopulation geneticsTrait lociImmune cell differentiationTranscriptomic landscapeTrajectory inferenceTranscriptional featuresInstrumental variablesCellular developmentProDualNet: dual-target protein sequence design method based on protein language model and structure model
Cheng L, Wei T, Cui X, Chen H, Yu Z. ProDualNet: dual-target protein sequence design method based on protein language model and structure model. Briefings In Bioinformatics 2025, 26: bbaf391. PMID: 40856523, PMCID: PMC12378908, DOI: 10.1093/bib/bbaf391.Peer-Reviewed Original ResearchConceptsSequence-structure informationProtein language modelsAllosteric bindingIn silico evaluationSequence design methodProtein sequencesProtein designBinding proteinBiological processesMulti-target strategyProteinHeterogeneous graph networkMultiple receptorsMultiple test setsLanguage modelGraph networkBindingExperimental structuresSingle receptorDesign networksReceptorsTest setTherapeutic potentialUnveiling fine-scale spatial structures and amplifying gene expression signals in ultra-large ST slices with HERGAST
Gong Y, Yuan X, Jiao Q, Yu Z. Unveiling fine-scale spatial structures and amplifying gene expression signals in ultra-large ST slices with HERGAST. Nature Communications 2025, 16: 3977. PMID: 40295488, PMCID: PMC12037780, DOI: 10.1038/s41467-025-59139-w.Peer-Reviewed Original ResearchConceptsGene expression signalsSpatial transcriptomics data analysisExpression signalsTranscriptome data analysisHeterogeneous graph networkReal-world datasetsSpatial expression patternsOver-smoothing problemSpatial transcriptomics dataGlobal spatial relationshipsST data analysisTranscriptome dataUltra-large-scaleConquer strategyExpression patternsGraph networkData splittingGenesIncreased diagnostic accuracy and better morphology characterization of unruptured intracranial aneurysm by ultra-high-resolution photon-counting detector CT angiography
He N, Lyu H, Zhang Y, Li R, Xu Z, Haacke E, Cui Y, Li J, Dong H, Han W, Chang R, Hu Z, Zhu C, Yu Z, Lu Y, Jiang H, Yan F. Increased diagnostic accuracy and better morphology characterization of unruptured intracranial aneurysm by ultra-high-resolution photon-counting detector CT angiography. Journal Of NeuroInterventional Surgery 2025, jnis-2025-023094. PMID: 40185624, DOI: 10.1136/jnis-2025-023094.Peer-Reviewed Original ResearchDigital subtraction angiographyUnruptured intracranial aneurysmsPer-aneurysm basisDiagnostic accuracyInter-rater agreementIntracranial aneurysmsIncrease diagnostic accuracyUIA managementCT angiographyProspective studyCerebral vascular diseaseSubtraction angiographyPatient managementWall calcificationPer-vesselAneurysmAngiographyVascular diseaseMorphological evaluationAneurysm irregularityUHRDetect unruptured intracranial aneurysmsUltra-high-resolution (UHRHypointensityReconstructed images