Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Disentangling Latent Risk Pathways via Bayesian Hypergraph Inference
Shengxian Ding, Haonan Gao, Pangpang Liu, Xinyuan Tian, Yize Zhao
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 repulsion prior (lambda>0) (ablation) | BHPI · Simulated data (N=2000) | test AUC x100: 74.69 -> 75.00 | No clear effect |
| add repulsion prior (lambda>0) (ablation) | BHPI · Simulated data (N=2000) | Jaccard stability of hyperedges: 0.542 -> 0.565 | Improved |
| add repulsion prior (lambda>0) (ablation) | BHPI · Simulated data (N=5000) | Jaccard stability of hyperedges: 0.581 -> 0.619 | Improved |
| add repulsion prior (lambda>0) (ablation) | BHPI · Simulated data (N=2000) | effective hyperedges per risk factor: 1.992 -> 1.874 | Improved |
| use hypergraph pathways instead of independent logistic | BHPI · Simulated data (N=2000) | test AUC x100: 73.86 -> 75.00 | Improved |
| use BHPI instead of LightGBM | BHPI · Simulated data (N=2000) | test AUC x100: 68.50 -> 75.00 | Improved |
| use BHPI for rare diseases (<2% prevalence) | BHPI vs logistic · UKB (prevalence <2%) | mean delta AUC x100: +1.2 | Improved |
| replace hypergraph with Bayesian factor regression K=30 | BFR · UKB (prevalence <2%) | mean delta AUC x100 vs logistic: -0.5 | Hurt |
| expand hyperedges into pairwise cliques | BHPI hyperedges · UKB (prevalence 2-5%) | mean delta AUC x100 vs logistic: +0.2 -> -0.2 | Hurt |
| keep only high-PIP edges (ablation) | BHPI · UKB (prevalence <2%) | mean AUC x100: 68.9 -> 68.7 | No clear effect |
| keep only low-PIP edges (ablation) | BHPI · Simulation | mean AUC x100: 75.7 -> 51.2 | Hurt |
| keep only a random subset of edges (ablation) | BHPI · Simulation | mean AUC x100: 75.7 -> 65.7 | Hurt |
Closest papers by meaning
- Risk Horizons: Structured Hypothesis Spaces for Longitudinal Clinical Prediction (NeurIPS 2026 · similarity 0.86)
- Electronic Health Record Analysis via Deep Poisson Factor Models (JMLR 2016 · similarity 0.86)
- Unscrambling disease progression at scale: fast inference of event permutations with optimal transport (NeurIPS 2024 · similarity 0.84)
- Exploiting Negative Samples: A Catalyst for Cohort Discovery in Healthcare Analytics (ICML 2024 · similarity 0.83)
- Hypergraph Representation Learning with Hyperlink Random Effects (NeurIPS 2026 · similarity 0.83)
- Multimodal Disease Progression Modeling via Spatiotemporal Disentanglement and Multiscale Alignment (NeurIPS 2025 · similarity 0.83)
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.
