Saved in:
Bibliographic Details
Main Authors: Liu, Chen, Yao, Wenfang, Yin, Kejing, Cheung, William K., Qin, Jing
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.11112
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915549569613824
author Liu, Chen
Yao, Wenfang
Yin, Kejing
Cheung, William K.
Qin, Jing
author_facet Liu, Chen
Yao, Wenfang
Yin, Kejing
Cheung, William K.
Qin, Jing
contents 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: (1) redundancy in consecutive CXR sequences, where static anatomical regions dominate over clinically-meaningful dynamics, and (2) temporal misalignment between sparse, irregular imaging and continuous EHR data. We introduce $\texttt{DiPro}$, 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 $\texttt{DiPro}$ could effectively extract temporal clinical dynamics and achieve state-of-the-art performance on both disease progression identification and general ICU prediction tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2510_11112
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Multimodal Disease Progression Modeling via Spatiotemporal Disentanglement and Multiscale Alignment
Liu, Chen
Yao, Wenfang
Yin, Kejing
Cheung, William K.
Qin, Jing
Computer Vision and Pattern Recognition
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: (1) redundancy in consecutive CXR sequences, where static anatomical regions dominate over clinically-meaningful dynamics, and (2) temporal misalignment between sparse, irregular imaging and continuous EHR data. We introduce $\texttt{DiPro}$, 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 $\texttt{DiPro}$ could effectively extract temporal clinical dynamics and achieve state-of-the-art performance on both disease progression identification and general ICU prediction tasks.
title Multimodal Disease Progression Modeling via Spatiotemporal Disentanglement and Multiscale Alignment
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2510.11112