Multimodal Disease Progression Modeling via Spatiotemporal Disentanglement and Multiscale Alignment

Chen Liu, Wenfang Yao, Kejing Yin, William K. Cheung, Jing Qin

Hong Kong Polytechnic University· Hong Kong Baptist University

Longitudinal Multimodal Fusion; Disease Progression Modeling; Clinical Prediction

⋅ NeurIPS ⋅ Project Page ⋅Slides ⋅Poster ⋅OpenReview

Abstract

Longitudinal multimodal data, including electronic health records (EHR) and sequential chest X-rays (CXRs), is critical for modeling disease progression, yet remains underutilized due to two key challenges:

redundancy in consecutive CXR sequences, where static anatomical regions dominate over clinically-meaningful dynamics, and
temporal misalignment between sparse, irregular imaging and continuous EHR data.

We introduce DiPro, a novel framework that addresses these challenges through region-aware disentanglement and multi-timescale alignment. First, we disentangle static (anatomy) and dynamic (pathology progression) features in sequential CXRs, prioritizing disease-relevant changes. Second, we hierarchically align these static and dynamic CXR features with asynchronous EHR data via local (pairwise interval-level) and global (full-sequence) synchronization to model coherent progression pathways.

Extensive experiments on the MIMIC dataset demonstrate that DiPro could effectively extract temporal clinical dynamics and achieve state-of-the-art performance on both disease progression identification and general ICU prediction tasks.