Research map / Probabilistic and scientific modeling / Dynamical systems
CLARITY: Medical World Model for Guiding Treatment Decisions by Simulating Context-Aware Disease Trajectories in Latent Space
Tianxingjian Ding, Yuanhao Zou, Chen Chen, Shah Mubarak, Yu Tian
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 |
|---|---|---|---|
| replace diffusion synthesis with latent dynamics (ablation) | CLARITY · MU-Glioma-Post | F1 (%): 43.6 -> 52.4 | Improved |
| add patient clinical context (ablation) | CLARITY latent · MU-Glioma-Post | F1 (%): 52.4 -> 52.8 | Improved |
| add inverse survival feedback (K=3) (ablation) | CLARITY latent · MU-Glioma-Post | F1 (%): 52.4 -> 54.0 | Improved |
| add context and feedback together (ablation) | CLARITY latent · MU-Glioma-Post | F1 (%): 52.4 -> 57.1 | Improved |
| increase feedback iterations from 1 to 3 (ablation) | Inverse survival evaluation · MU-Glioma-Post | F1 (%): 53.1 -> 57.1 | Improved |
| increase feedback iterations from 3 to 4 (ablation) | Inverse survival evaluation · MU-Glioma-Post | F1 (%): 57.1 -> 55.4 | Hurt |
| add Brier loss (ablation) | Cox loss · MU-Glioma-Post | C-index (%): 76.1 -> 76.8 | Improved |
| add hard contrastive loss (ablation) | Brier + Cox · MU-Glioma-Post | C-index (%): 76.8 -> 77.3 | Improved |
| use soft contrastive labels instead of hard (ablation) | contrastive loss · MU-Glioma-Post | C-index (%): 77.3 -> 79.4 | Improved |
| use soft instead of hard contrastive labels (ablation) | contrastive loss · MU-Glioma-Post | F1 (%): 57.1 -> 55.4 | Hurt |
| use MRI-CORE with LoRA instead of DINOv2 (ablation) | visual encoder · MU-Glioma-Post | C-index: 0.665 -> 0.794 | Improved |
| use MedSigLIP instead of DINOv2 (ablation) | visual encoder · MU-Glioma-Post | C-index: 0.665 -> 0.705 | Improved |
Closest papers by meaning
- Medical World Model (ICCV 2025 · similarity 0.90)
- Medical LLMs as Medical World Models: Unified Policy-Dynamics Learning with Test-Time Search (NeurIPS 2026 · similarity 0.89)
- MTBBench: A Multimodal Sequential Clinical Decision-Making Benchmark in Oncology (NeurIPS 2025 · similarity 0.87)
- Learning Patient-Specific Disease Dynamics With Latent Flow Matching For Longitudinal Imaging Generation (ICLR 2026 · similarity 0.87)
- Contextualized Policy Recovery: Modeling and Interpreting Medical Decisions with Adaptive Imitation Learning (ICML 2024 · similarity 0.85)
- EpiCare: A Reinforcement Learning Benchmark for Dynamic Treatment Regimes (NeurIPS 2024 · 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.
