# Clin-JEPA: A Multi-Phase Co-Training Framework for Joint-Embedding Predictive Pretraining on EHR Patient Trajectories

arXiv: https://arxiv.org/abs/2605.10840 · HTML: https://mlsanity.com/paper/2605.10840/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Medical and surgical world models, one of 11 directions in World models.
- Builds on 12 of its 38 references that are in the map.
- First posted on arXiv 11 May 2026.

## Placed in

- World models / [Medical and surgical world models](https://mlsanity.com/wm/medical-and-surgical-world-models/)

## Builds on (most cited first, 12 of 12)

- [An Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling](https://mlsanity.com/paper/1803.01271/) (2018)
- [V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning](https://mlsanity.com/paper/2506.09985/) (2025)
- [Revisiting Feature Prediction for Learning Visual Representations from Video](https://mlsanity.com/paper/2404.08471/) (2024)
- [VL-JEPA: Joint Embedding Predictive Architecture for Vision-language](https://mlsanity.com/paper/2512.10942/) (2025)
- [LLM-JEPA: Large Language Models Meet Joint Embedding Predictive Architectures](https://mlsanity.com/paper/2509.14252/) (2025)
- [EHRMamba: Towards Generalizable and Scalable Foundation Models for Electronic Health Records](https://mlsanity.com/paper/2405.14567/) (2024)
- [EHR-R1: A Reasoning-Enhanced Foundational Language Model for Electronic Health Record Analysis](https://mlsanity.com/paper/2510.25628/) (2025)
- [Foundation models for electronic health records: representation dynamics and transferability](https://mlsanity.com/paper/2504.10422/) (2025)
- [Beyond Generative AI: World Models for Clinical Prediction, Counterfactuals, and Planning](https://mlsanity.com/paper/2511.16333/) (2025)
- [Building the EHR Foundation Model via Next Event Prediction](https://mlsanity.com/paper/2509.25591/) (2025)
- [CLARITY: Medical World Model for Guiding Treatment Decisions by Simulating Context-Aware Disease Trajectories](https://mlsanity.com/paper/2512.08029/) (2025)
- [The Patient is not a Moving Document: A World Model Training Paradigm for Longitudinal EHR](https://mlsanity.com/paper/2601.22128/) (2026)

## Built on by (most cited first, 2 of 2)

- [The JEPA Predictor: A Transferable Operator for Occluded Feature Completion](https://mlsanity.com/paper/2607.16274/) (2026)
- [Intervention Granularity Matters: Coherent Treatment Bundles in Counterfactual Simulation with Clinical World Models](https://mlsanity.com/paper/2609.21906/) (2026)

## Abstract (arXiv)

Joint-embedding predictive architectures (JEPA) learn representations by predicting in latent space, as in computer vision; retaining the action-conditioned predictor at inference turns them into latent world models, enabling planning in robotics (V-JEPA 2-AC). Bringing this design to EHR patient trajectories---a predictor that simulates a patient's trajectory in latent space---has not been explored. We use an LLM as the encoder, reading the hourly record as text, avoiding feature engineering and vocabulary harmonisation. But an LLM adapted by supervised fine-tuning does not organise its latent space around physiological dynamics, and freezing it to train the predictor, as in V-JEPA 2-AC, leaves the encoder unaware of the rollout signal: the predictor degrades under rollout. We instead co-train encoder and predictor under one latent-prediction objective, grounding the encoder in the dynamics its predictor must follow. Na"ive co-training, however, is unstable: the untrained predictor drags the encoder toward collapse, and the predictor's rollout diverges as its target space moves. We present Clin-JEPA, a five-phase curriculum that stably co-trains an LLM encoder with a latent trajectory predictor on MIMIC-IV. Three evaluations support the design: (1) under 48-hour autoregressive rollout the co-trained predictor degrades least (predictor degradation ×1.06, against ×1.23--1.36 for two-stage designs and ×6.3--66 for curriculum ablations) while the co-trained encoder resolves the progression of patient state most sharply (largest state displacement); (2) the co-trained encoder separates deteriorating from stable patients in its latent space with Cohen's d=1.59, against ≤0.50 for two-stage encoders; (3) one set of embeddings serves 34 downstream tasks across three benchmarks, outperforming strong per-task tuned baselines and a pretrained EHR foundation model.