PulmoWorld models pneumonia progression from mixed clinical data
PulmoWorld: An Asynchronous Multimodal Medical World Model for Longitudinal Pneumonia Dynamics
Computer Vision and Pattern Recognition
Summary
Pneumonia affects different parts of the body in complex ways, and measurements about it often come at different times and may be incomplete. The authors present PulmoWorld, a model that learns to track and predict pneumonia’s changes over time by updating its beliefs only when new data arrives. This lets it make forecasts about future health states and suggest safe treatment plans. Tests show it performs well in predicting patient outcomes and supports better decision-making in simulated settings.
What this means in practice
- •For hospital data teams: Improve pneumonia patient monitoring by integrating irregular and partial clinical data for better prognosis and treatment planning.
- •For medical device integrators: Incorporate asynchronous clinical measurements into predictive models that simulate pneumonia progression and support automated decision tools.
Authors
Cheng Tang, Xinyu Zhang, Jiawei Chen, Minghui Zhang, Shaocong Wei, Wei Li, Yuxing Chen, Kongcheng Zhang, Chun-Mei Feng, Nanqing Dong, Xinyuan Zhu, Ziyan Huang, Junjun He, Yirong Chen, Lingtong Huang, Min Cen, Ming Hu
Abstract
Pneumonia changes across respiration, inflammation, and organ function, while the measurements that describe it arrive at different times and are often incomplete. We introduce PULMOWORLD, a world model for longitudinal pneumonia dynamics that connects asynchronous state estimation, recursive simulation, and evidence-constrained planning. History-aware filtering separates state propagation from observation-driven correction, so a measurement updates the shared belief only when it is observed. That belief supports recursive forecasts of clinical values and modality representations, and summaries of future prognosis. Cap-MBRL searches imagined continuations inside an authorized action set and replaces a frozen fallback only when a candidate meets the risk budget and a predicted-cost margin. On the pneumonia cohort MASS, recursive forecasts are competitive with task-trained sequence models, and simulated future states retain prognostic information. Trained separately on public MIMIC-IV-CXR chest radiographs and clinical trajectories, this model improves prognostic discrimination over latest-value fusion in development and is competitive on long-horizon error on the locked pneumonia panel. In five known-mechanism MASS-Syn environments, Cap-MBRL lowers selection regret relative to direct fallback execution.