Research map / Vision-language and multimodal / Cross-modal learning
Factorized Context Aggregation for Robust Cancer Risk Estimation via Soft Re-Ranked Retrieval and Hierarchical Anchors
Puria Azadi Moghadam, Ali Khajegili Mirabadi, Behnam Maneshgar, Hossein Farahani, Ali Bashashati
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add soft re-ranking of retrieved cases (ablation) | histology-only baseline · TCGA overall | C-index: 0.569 -> 0.593 | Improved |
| add factorized prognostic context aggregation (ablation) | soft re-ranking model · TCGA overall | C-index: 0.593 -> 0.607 | Improved |
| add intra-modality anchor loss (ablation) | retrieval + aggregation model · TCGA overall | C-index: 0.607 -> 0.609 | Improved |
| add inter-modality anchor loss (ablation) | retrieval + aggregation model · TCGA overall | C-index: 0.607 -> 0.609 | Improved |
| add both intra- and inter-modality anchors (ablation) | retrieval + aggregation model · TCGA overall | C-index: 0.607 -> 0.617 | Improved |
| replace histology-only model with proposed method | histology baseline · TCGA overall | C-index: 0.569 -> 0.617 | Improved |
| compare with fully multimodal model at test | proposed method · TCGA overall | C-index: 0.626 -> 0.617 | Hurt |
| increase training missing rate of auxiliary modalities up to 40% (ablation) | proposed method · TCGA | C-index | No clear effect |
| use large retrieval depth K for report retrieval (ablation) | proposed method · TCGA H+R | C-index | Hurt |
| use small retrieval depth K (ablation) | proposed method · TCGA | C-index | Hurt |
| use higher K for gene-expression retrieval (ablation) | proposed method · TCGA H+G | C-index | Improved |
| use proposed method instead of best SOTA for stratification | proposed method · TCGA stratification | significant settings: 12 -> 16 of 24 | Improved |
Closest papers by meaning
- Interpretable Vision-Language Survival Analysis with Ordinal Inductive Bias for Computational Pathology (ICLR 2025 · similarity 0.86)
- Modeling Dense Multimodal Interactions Between Biological Pathways and Histology for Survival Prediction (CVPR 2024 · similarity 0.86)
- Prototypical Information Bottlenecking and Disentangling for Multimodal Cancer Survival Prediction (ICLR 2024 · similarity 0.86)
- Multimodal Prototyping for cancer survival prediction (ICML 2024 · similarity 0.85)
- Robust Multimodal Survival Prediction with Conditional Latent Differentiation Variational AutoEncoder (CVPR 2025 · similarity 0.85)
- Risk Horizons: Structured Hypothesis Spaces for Longitudinal Clinical Prediction (NeurIPS 2026 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
