AI summaryⓘ
The authors created a new method to predict patient recovery after medical procedures by looking at all kinds of information collected at different times, not just at the start and end. Their model uses initial brain scans to build a 3D snapshot of the patient’s condition, then updates this snapshot as new medical events and measurements happen over time. They tested this approach on patients recovering from a heart procedure called atrial fibrillation ablation and found it predicted the chance of the problem coming back fairly well. The model can also estimate how much scarring occurred without needing extra scans later on. This helps doctors understand risks over time and adjust predictions as new data comes in.
clinical prediction modelslatent stateatrial fibrillation ablation3D imagingpost-intervention recoverytime-ordered eventsrisk assessmentAUROCAUPRCmean absolute error (MAE)
Authors
Yunsung Chung, Yingshuo Liu, Abboud F. Hassan, Han Feng, Mary M. Maleckar, Nassir Marrouche, Jihun Hamm
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.