Intervention-Aware Clinical World Model for Post-Op Outcome Forecasting in Cardiology

2026-08-13

AI Summary

Traditional clinical prediction models oversimplify postoperative outcomes as static mappings, neglecting the impact of asynchronous interventions and dynamic physiological changes during recovery. This work proposes the first intervention-aware clinical world model that formulates postoperative recovery as a temporal process driven by asynchronous clinical events. By encoding a 3D latent state that integrates baseline imaging, surgical context, static covariates, and peri-event physiological signals, the model enables dynamic state evolution over time. It supports multi-temporal risk queries and retrospective input editing, and can predict scar burden without requiring follow-up MRI. Evaluated on the DECAAF-II dataset, the model achieves an AUROC of 0.756 and AUPRC of 0.777 for predicting atrial fibrillation recurrence at 90 days post-ablation, with a mean absolute error of 2.971 percentage points in scar extent prediction.

Abstract

Many clinical prediction models treat post-intervention outcomes as a one-step mapping from baseline measurements to a future endpoint. However, recovery after a procedure often unfolds as an irregular trajectory: clinical observations, medication changes, repeat interventions, and physiological measurements are recorded asynchronously and can change risk assessment over time. We propose an intervention-aware clinical world model that represents each patient with a structured latent state and evolves it through time-ordered post-intervention events. The model first encodes baseline imaging into a 3D spatial latent state. It then updates this state using procedural context, static covariates, elapsed time, and peri-event physiological embeddings. Follow-up imaging provides training-only supervision through a latent forecasting objective. We apply the framework to atrial fibrillation ablation. During the 90-day recovery window, irregular post-procedure records provide clinically meaningful evidence for long-term recurrence risk. In repeated internal cross-validation on DECAAF-II, our model achieves AUROC 0.756 and AUPRC 0.777 for recurrence prediction. It also achieves a scar-extent MAE of 2.971 percentage points without requiring follow-up MRI intensities at inference. The learned state supports recurrence-risk queries at different horizons and retrospective input editing of blanking-period records.

Problem

• post-intervention outcome forecasting
• irregular clinical trajectories
• dynamic risk assessment
• atrial fibrillation ablation
• clinical world modeling

Innovation

• intervention-aware modeling
• latent state dynamics
• post-operative trajectory forecasting
• 3D spatial latent representation
• clinical world model