World models for biomedicine.

Biomedical world models spanning molecular to clinical scales can predict system responses to interventions, enabling proactive decision-making in medicine.

• Why it matters: Understanding dynamic biological responses is crucial for improving interventions and personalized treatments, yet current models lack comprehensive, multi-scale predictive capabilities.
• What they did: The authors propose developing integrated world models that combine observational, interventional, and longitudinal data across modalities to simulate biological trajectories.
• The result: These models could revolutionize biomedical research by forecasting responses to novel perturbations, identifying effective interventions, and guiding experimental planning over extended periods.