Survival Prediction on TCGA-UCEC
0.783C-indexSTEPH
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| STEPHTraining Source=Model merging-based knowledge transfer from Ms2026.03 | 0.783 | — | — | — | |
| HFGPIG. (Genomic)=✓, P. (Proteomic)=✓, I. (Pathology Imaging)=✓2026.03 | 0.782 | — | — | — | |
| DeltaMILFeature Extractor=UNI [7]2025.12 | 0.7713 | — | — | — | |
| DeltaMILFeature Extractor=UNI2025.12 | 0.7713 | — | — | — | |
| DeltaMILFeature Extractor=ResNet-502025.12 | 0.7712 | — | — | — | |
| ABMILFeature Extractor=UNI [7]2025.12 | 0.7702 | — | — | — | |
| ABMILFeature Extractor=UNI2025.12 | 0.7702 | — | — | — | |
| MUSTPathology modality (P)=Observed, Genomic modality (G)=Observed2026.03 | 0.768 | — | — | — | |
| DPsurvUncertainty-Aware=true, Validation=5-fold cross-validation2025.09 | 0.766 | 0.251 | 0.692 | — | |
| PANTHERAll + MLPLearning Paradigm=Ours, Features=UNI, Prototypes (C)=16, Evaluation Protocol=MLP2024.05 | 0.757 | — | — | — | |
| PS3†G. (Genomic)=✓, P. (Proteomic)=✓, I. (Pathology Imaging)=✓2026.03 | 0.757 | — | — | — | |
| MUSTPathology modality (P)=Observed, Genomic modality (G)=Missing2026.03 | 0.755 | — | — | — | |
| PANTHERWA + lin.Learning Paradigm=Ours, Features=UNI, Prototypes (C)=16, Evaluation Protocol=Linear2024.05 | 0.753 | — | — | — | |
| MambaMILFeature Extractor=ResNet-502025.12 | 0.7526 | — | — | — | |
| Mamba2MILFeature Extractor=ResNet-502025.12 | 0.7524 | — | — | — | |
| MurreNetPathological images=true, Genomic profiles=true2025.07 | 0.752 | — | — | — | |
| PANTHERAll + lin.Learning Paradigm=Ours, Features=UNI, Prototypes (C)=16, Evaluation Protocol=Linear2024.05 | 0.751 | — | — | — | |
| HDMILGCE=false, Cross-validation=5-fold2026.05 | 0.75 | — | — | — | |
| MUSTPathology modality (P)=Missing, Genomic modality (G)=Observed2026.03 | 0.748 | — | — | — | |
| Mean-PoolingFeature Extractor=UNI [7]2025.12 | 0.7473 | — | — | — | |
| Mean-PoolingFeature Extractor=UNI2025.12 | 0.7473 | — | — | — | |
| OTLearning Paradigm=Unsupervised, Features=UNI, Prototypes (C)=162024.05 | 0.747 | — | — | — | |
| ShaSpecPathology modality (P)=Observed, Genomic modality (G)=Observed2026.03 | 0.747 | — | — | — | |
| CLAM-SBGCE=false, Cross-validation=5-fold2026.05 | 0.746 | — | — | — | |
| S4MILFeature Extractor=ResNet-502025.12 | 0.7452 | — | — | — | |
| MurreNet w/ IC-DiTTrain Data=Real+Synth2026.03 | 0.7415 | — | — | — | |
| MCATPathological images=true, Genomic profiles=true2025.07 | 0.74 | — | — | — | |
| SMILPathology modality (P)=Observed, Genomic modality (G)=Observed2026.03 | 0.74 | — | — | — | |
| ICFNet†G. (Genomic)=✓, P. (Proteomic)=✓, I. (Pathology Imaging)=✓2026.03 | 0.739 | — | — | — | |
| ROUPKTTraining Source=Representation-based knowledge transfer from Ms2026.03 | 0.7371 | — | — | — | |
| MOTCatPathological images=true, Genomic profiles=true2025.07 | 0.737 | — | — | — | |
| SurvPathPathological images=true, Genomic profiles=true2025.07 | 0.737 | — | — | — | |
| BayesMILUncertainty-Aware=true, Validation=5-fold cross-validation2025.09 | 0.736 | 0.679 | 1.941 | — | |
| MurreNet w/ PathDiffTrain Data=Real+Synth2026.03 | 0.7347 | — | — | — | |
| LD-CVAEPathology modality (P)=Observed, Genomic modality (G)=Observed2026.03 | 0.734 | — | — | — | |
| Iso-C AMTraining Source=Model merging-based knowledge transfer from Ms2026.03 | 0.7334 | — | — | — | |
| Mamba2MILFeature Extractor=UNI [7]2025.12 | 0.7325 | — | — | — | |
| Mamba2MILFeature Extractor=UNI2025.12 | 0.7325 | — | — | — | |
| MurreNetTrain Data=Real2026.03 | 0.7319 | — | — | — | |
| TransMILFeature Extractor=UNI [7]2025.12 | 0.7309 | — | — | — | |
| TransMILFeature Extractor=UNI2025.12 | 0.7309 | — | — | — | |
| DeepSetsLearning Paradigm=Unsupervised, Features=UNI2024.05 | 0.73 | — | — | — | |
| MambaMILFeature Extractor=UNI [7]2025.12 | 0.7296 | — | — | — | |
| MambaMILFeature Extractor=UNI2025.12 | 0.7296 | — | — | — | |
| DSMILPathological images=true, Genomic profiles=false2025.07 | 0.729 | — | — | — | |
| ShaSpecPathology modality (P)=Observed, Genomic modality (G)=Missing2026.03 | 0.727 | — | — | — | |
| S4MILFeature Extractor=UNI [7]2025.12 | 0.7268 | — | — | — | |
| S4MILFeature Extractor=UNI2025.12 | 0.7268 | — | — | — | |
| LD-CVAEPathology modality (P)=Observed, Genomic modality (G)=Missing2026.03 | 0.726 | — | — | — | |
| ASMILFeatures=CONCH2026.03 | 0.7243 | — | — | — | |
| DTFD-MILPathological images=true, Genomic profiles=false2025.07 | 0.724 | — | — | — | |
| Patch-GCNFeatures=CONCH2026.03 | 0.7212 | — | — | — | |
| TransMILGCE=false, Cross-validation=5-fold2026.05 | 0.721 | — | — | — | |
| CMTA w/ IC-DiTTrain Data=Real+Synth2026.03 | 0.7208 | — | — | — | |
| TIES AMTraining Source=Model merging-based knowledge transfer from Ms2026.03 | 0.7174 | — | — | — | |
| MCATG. (Genomic)=✓, I. (Pathology Imaging)=✓2026.03 | 0.716 | — | — | — | |
| ILRAUncertainty-Aware=false, Validation=5-fold cross-validation2025.09 | 0.716 | 0.92 | 4.074 | — | |
| TransMILFeature Extractor=ResNet-502025.12 | 0.7155 | — | — | — | |
| Porpoise w/ IC-DiTTrain Data=Real+Synth2026.03 | 0.7151 | — | — | — | |
| H2TLearning Paradigm=Unsupervised, Features=UNI, Prototypes (C)=162024.05 | 0.715 | — | — | — | |
| Fine-tunedTraining Source=Representation-based knowledge transfer from Ms2026.03 | 0.7146 | — | — | — | |
| MoMEG. (Genomic)=✓, I. (Pathology Imaging)=✓2026.03 | 0.714 | — | — | — | |
| Fine-tunedTraining Source=Traditional cancer-specific training on Mt2026.03 | 0.7136 | — | — | — | |
| MHIM-MILGCE=false, Cross-validation=5-fold2026.05 | 0.712 | — | — | — | |
| VanillaTraining Source=Traditional cancer-specific training on Mt2026.03 | 0.7098 | — | — | — | |
| M3CarePathology modality (P)=Observed, Genomic modality (G)=Observed2026.03 | 0.709 | — | — | — | |
| PANTHERUncertainty-Aware=false, Validation=5-fold cross-validation2025.09 | 0.709 | 0.866 | 3.045 | — | |
| DTFD-MILGCE=false, Cross-validation=5-fold2026.05 | 0.706 | — | — | — | |
| ResTopoMIL2026.05 | 0.7058 | — | — | — | |
| CMTA w/ PathDiffTrain Data=Real+Synth2026.03 | 0.7052 | — | — | — | |
| TransMILPathological images=true, Genomic profiles=false2025.07 | 0.705 | — | — | — | |
| CMTAPathological images=true, Genomic profiles=true2025.07 | 0.705 | — | — | — | |
| MOTCatG. (Genomic)=✓, I. (Pathology Imaging)=✓2026.03 | 0.705 | — | — | — | |
| SNNPathological images=false, Genomic profiles=true2025.07 | 0.703 | — | — | — | |
| PANTHERTop + lin.Learning Paradigm=Ours, Features=UNI, Prototypes (C)=16, Evaluation Protocol=Linear2024.05 | 0.703 | — | — | — | |
| SMILPathology modality (P)=Missing, Genomic modality (G)=Observed2026.03 | 0.703 | — | — | — | |
| CMTAG. (Genomic)=✓, I. (Pathology Imaging)=✓2026.03 | 0.702 | — | — | — | |
| PIBDG. (Genomic)=✓, I. (Pathology Imaging)=✓2026.03 | 0.702 | — | — | — | |
| 2DMambaMIL2026.05 | 0.702 | — | — | — | |
| CAMILGCE=false, Cross-validation=5-fold2026.05 | 0.702 | — | — | — | |
| ABMILPathological images=true, Genomic profiles=false2025.07 | 0.701 | — | — | — | |
| Porpoise w/ PathDiffTrain Data=Real+Synth2026.03 | 0.6997 | — | — | — | |
| Mean-PoolingFeature Extractor=ResNet-502025.12 | 0.6994 | — | — | — | |
| ABMILFeature Extractor=ResNet-502025.12 | 0.699 | — | — | — | |
| TransMILUncertainty-Aware=false, Validation=5-fold cross-validation2025.09 | 0.698 | 0.936 | 4.59 | — | |
| DGR-MIL2026.05 | 0.6976 | — | — | — | |
| CMTATrain Data=Real2026.03 | 0.6975 | — | — | — | |
| Model Avg.Training Source=Model merging-based knowledge transfer from Ms2026.03 | 0.6971 | — | — | — | |
| Surgery AMTraining Source=Model merging-based knowledge transfer from Ms2026.03 | 0.6964 | — | — | — | |
| ABMILGCE=false, Cross-validation=5-fold2026.05 | 0.696 | — | — | — | |
| MurreNet w/ IC-DiTTrain Data=Synth2026.03 | 0.6959 | — | — | — | |
| TransMILLearning Paradigm=Supervised, Features=UNI2024.05 | 0.695 | — | — | — | |
| PorpoiseTrain Data=Real2026.03 | 0.6918 | — | — | — | |
| MMPTrans.G. (Genomic)=✓, I. (Pathology Imaging)=✓2026.03 | 0.69 | — | — | — | |
| M3CarePathology modality (P)=Observed, Genomic modality (G)=Missing2026.03 | 0.69 | — | — | — | |
| DSMILUncertainty-Aware=false, Validation=5-fold cross-validation2025.09 | 0.69 | 0.652 | 1.871 | — | |
| DeReFGenomic=true, Pathology=true2025.08 | 0.688 | — | — | — | |
| SurvPathG. (Genomic)=✓, I. (Pathology Imaging)=✓2026.03 | 0.688 | — | — | — | |
| CCLGenomic=true, Pathology=true2025.08 | 0.686 | — | — | — | |
| MOMEPathological images=true, Genomic profiles=true2025.07 | 0.685 | — | — | — |