Summary
Predicting how a patient's medical images will change over time helps doctors track diseases like breast cancer. The authors point out that just being accurate isn’t enough — we need to know if the prediction truly relies on that patient’s own history. They propose a way, called P³ (Patient, Place, Prior), to check if the forecast actually uses personal patient data beyond general trends. Their method shows that including a patient’s specific past images and spatial information can reduce errors, but sometimes the improvement over average predictions is not statistically clear. This approach helps clarify what counts as truly personalized predictions in medical imaging.
What this means in practice
- •For medical imaging teams: Use patient-specific past imaging and spatial lesion data to improve breast cancer MRI forecast accuracy during treatment.
- •For healthcare data analysts: Evaluate medical image prediction models to verify if they truly rely on individual patient data instead of just population averages.
Tested on one dataset.
Authors
Xingrui Gu, Hanxue Gu, Yuxiang Zhang, Yang Yang
Abstract
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.