JEPA · 2026-10-08
Patient, Place, Prior (P$^3$): What Counts as Personalization in Medical World Models?
Xingrui Gu, Hanxue Gu, Yuxiang Zhang, Yang Yang
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
The authors introduce Patient, Place, Prior (P^3), an audit framework to separate (1) whether a longitudinal medical forecast benefits from a patient's own history (Patient), (2) whether it benefits from patient-matched externally supplied spatial support (Place), and (3) whether it adds value beyond a population-average under matched support and context (Prior). They instantiate it with Cancer JEPA, a lesion-constrained one-step latent forecasting model.
Why it matters
Accuracy alone cannot establish that a medical world model is truly personalized. P^3 audits predictive attribution by varying patient history and externally supplied spatial support independently at inference, while keeping the future target and evaluation procedure fixed. Applied to Cancer JEPA on longitudinal breast DCE-MRI, the authors report lower forecast error when using a patient's own history and patient-matched lesion occupancy maps, while patient-specific predictive value beyond a population-level pattern remains unresolved.
Method
- Define P^3 contrasts: Patient compares patient history vs another patient’s history under patient-matched support; Place compares patient-matched support vs substituted support under patient history; Prior compares patient-history prediction vs a population-average prediction under matched support and observed context.
- Use a fixed longitudinal predictor and vary only the source of history (C) and/or externally supplied spatial support (S) at inference; model parameters, forecast horizon/context, and future target/evaluation procedure remain fixed.
- Instantiate P^3 with Cancer JEPA: frozen encoder representations, a low-complexity reduced-rank regression (RRR) baseline plus an occlusion-trained lesion-constrained neural correction gated by supplied core occupancy.
Limitation
The authors state that all P^3 analyses are post-hoc and descriptive because the validation cohort had already been used during model development, and the present experiments neither access nor report the test partition.
Abstract (from arXiv)
Longitudinal models forecast how a patient's imaging state evolves, but accuracy does not show whether the patient's observed trajectory drives the prediction. A population-average forecast may be useful but cannot establish a patient-specific world-model claim. We introduce Patient, Place, Prior (P$^3$), an audit asking whether a forecast benefits from the patient's longitudinal imaging history (Patient), benefits from patient-matched externally supplied spatial support (Place), and gains predictive value beyond a population-average prediction under matched support and context (Prior). We also propose Cancer JEPA, a one-step model that forecasts frozen representations of future breast dynamic contrast-enhanced MRI examinations during neoadjuvant therapy. It adds a lesion-constrained neural correction, trained with an occlusion-based latent objective, to a patient-conditioned low-complexity reduced-rank regression baseline. This factorization permits a post-hoc P$^3$ audit of the frozen model. In a validation cohort previously used in development, forecast error is lower when the neural correction receives the patient's history rather than another patient's and patient-matched lesion occupancy maps rather than substituted maps. However, the descriptive 95% interval comparing the correction computed from patient history with the population-average neural correction includes zero. P$^3$ thus separates input use from evidence of patient-specific predictive value beyond a population-level pattern.
Related papers
- Repeated-Measure Leakage, Distribution Shift, and Reliability under Partial Observation in Patient World Models
- PatientWM: Towards Generalizable Clinical World Models via Unified Patient States
- Forecasting Patient Trajectories from the Electronic Health Records with Action-Conditioned World Models
- CoRe-WM: Conditioned Response-State World Modeling for Longitudinal Treatment-Response Prediction in Breast DCE-MRI
- HyperNSDE: Personalized Neural SDEs for Joint Static-Longitudinal Clinical Data Generation
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First public release: 2026-10-06 · Indexed by Pub-AI: 2026-10-08 UTC
Patient, Place, Prior (P$^3$): What Counts as Personalization in Medical World Models?
Preprint posted: 2026-10-06 · Original record ↗
- arxiv
- 2610.09194
Explicit source versions (1)
- v1 · 2026-10-06
- Preprint available
Source date: 2026-10-06 · First detected: 2026-10-25 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · atom:published (official metadata)
Abstract · arxiv
Longitudinal models forecast how a patient's imaging state evolves, but accuracy does not show whether the patient's observed trajectory drives the prediction. A population-average forecast may be useful but cannot establish a patient-specific world-model claim. We introduce Patient, Place, Prior (P$^3$), an audit asking whether a forecast benefits from the patient's longitudinal imaging history (Patient), benefits from patient-matched externally supplied spatial support (Place), and gains predictive value beyond a population-average prediction under matched support and context (Prior). We also propose Cancer JEPA, a one-step model that forecasts frozen representations of future breast dynamic contrast-enhanced MRI examinations during neoadjuvant therapy. It adds a lesion-constrained neural correction, trained with an occlusion-based latent objective, to a patient-conditioned low-complexity reduced-rank regression baseline. This factorization permits a post-hoc P$^3$ audit of the frozen model. In a validation cohort previously used in development, forecast error is lower when the neural correction receives the patient's history rather than another patient's and patient-matched lesion occupancy maps rather than substituted maps. However, the descriptive 95% interval comparing the correction computed from patient history with the population-average neural correction includes zero. P$^3$ thus separates input use from evidence of patient-specific predictive value beyond a population-level pattern.