ChronoMedicalWorld: A Medical World Model for Learning Patient Trajectories from Longitudinal Care Data

Long-horizon clinical simulation -- predicting how a patient's physiology evolves over years under specified interventions -- is central to chronic-disease care, yet existing electronic health record (EHR) models are predominantly discriminative, and general-purpose large language models drift under repeated interventions.

Medigy Insights

Researchers developed an AI “medical world model” that can simulate how a patient’s health may change over years under different treatments and interventions, outperforming a structured GPT-based baseline in forecasting chronic kidney disease progression. The model also incorporates clinician-patient communications, showing that conversations can provide valuable signals for predicting long-term health outcomes.

Continue reading at arxiv.org