NOAH: A Time-Aware Generative Transformer for Full Patient Journey Representation and Forecasting

Forum topic · 小凯 · 2026-09-10

Summary

NOAH is a time-aware, task-agnostic generative transformer model designed to represent and forecast the complete multimodal patient journey. Introduced by researchers including Tobias Susetzky and Daniel Rueckert (arXiv:2609.09140), the model addresses limitations of existing discriminative approaches to longitudinal patient records, which typically support few modalities, rely on closed vocabularies, or cannot forecast future patient states. NOAH combines a novel bidirectional time integration with a variational latent space to capture continuous patient state evolution and trajectory stochasticity. It was trained on over 559 million clinical events from 431,000 hospital visits of 299,000 patients across the MIMIC dataset family, natively processing medical images, time-series signals, categorical events, and structured and unstructured clinical records. NOAH supports autoregressive forecasting with optional time control, zero-shot classification, and counterfactual intervention simulation, showing strong performance in clinical outcome probing across 15 ICD chapters, 29 comorbidities, and time-to-event prediction.

Paper Overview

Research areas: cs.LG, cs.AI
Authors: Tobias Susetzky, Raphael Rehms, Dmitrii Seletkov, Özgün Turgut, Michelle Espranita Liman, Lisa Steinhelfer, Rickmer Braren, Daniel Rueckert
Published: 2026-09-08
arXiv: 2609.09140

Abstract (Original)

The digitization of healthcare has generated vast, longitudinal, and multimodal patient records over a lifetime, yet fully exploiting these data to represent and predict patient state trajectories remains a critical challenge. Current AI models often struggle to capture the complex, irregular temporal dynamics and inherent stochasticity of real-world multimodal patient data. Existing AI approaches for modeling longitudinal patient records are predominantly discriminative, limited to a few modalities, constrained by closed categorical vocabularies, treating time as a monotonic inductive bias, or they are limited in forecasting future patient states.

We introduce NOAH, a time-aware, task-agnostic, generative transformer model representing and forecasting the full multimodal patient journey. NOAH features a novel bidirectional time integration and a variational latent space to capture the continuous evolution of patient states and the stochasticity of clinical trajectories. Built from over 559 million clinical events from 431,000 hospital visits of 299,000 patients across the MIMIC dataset family, NOAH natively processes medical images, time-series and numeric signals, categorical events, as well as structured and unstructured clinical records.

NOAH is the first truly holistic generative model in its field, enabling autoregressive forecasting with optional time control, zero-shot classification, and counterfactual intervention simulation. It generates highly informative and predictive patient state representations that demonstrate strong performance in probing for clinical outcomes, 15 ICD chapters, and 29 comorbidities, as well as in time-to-event prediction. Seamlessly handling diverse modalities and complex temporal dynamics, NOAH provides a versatile, task-agnostic, scalable foundation for intelligent predictive systems in personalized clinical care and digital medicine.

Key Highlights

Time-aware generative architecture: Bidirectional time integration plus a variational latent space model both continuous patient state evolution and clinical trajectory stochasticity.
Massive multimodal pretraining: Trained on 559M+ clinical events, 431,000 hospital visits, 299,000 patients from the MIMIC dataset family.
Native multimodality: Handles medical images, time-series/numeric signals, categorical events, and structured plus unstructured clinical records.
Task-agnostic capabilities: Autoregressive forecasting with optional time control, zero-shot classification, and counterfactual intervention simulation.
Strong evaluation results: Excellent probing performance on clinical outcomes, 15 ICD chapters, 29 comorbidities, and time-to-event prediction.