Skip to main navigation Skip to search Skip to main content

Multimodal Disease Progression Modeling via Spatiotemporal Disentanglement and Multiscale Alignment

Research output: Chapter in book/report/conference proceeding › Conference proceeding › peer-review

2 Citations (Scopus)

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: (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 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.
Original languageEnglish
Title of host publication39th Conference on Neural Information Processing Systems, NeurIPS 2025
EditorsD. Belgrave, C. Zhang, H. Lin, R. Pascanu, P. Koniusz, M. Ghassemi, N. Chen
PublisherNeural Information Processing Systems Foundation
Pages146893-146920
Number of pages28
DOIs
Publication statusPublished - 2 Dec 2025
Event39th Conference on Neural Information Processing Systems, NeurIPS 2025 - San Diego, United States
Duration: 2 Dec 2025 → 7 Dec 2025
https://neurips.cc/Conferences/2025 (Conference website)
https://neurips.cc/virtual/2025/loc/san-diego/papers.html (Conference schedule)
https://proceedings.neurips.cc/paper_files/paper/2025 (Conference proceedings)

Publication series

NameAdvances in Neural Information Processing Systems
Volume38
ISSN (Print)1049-5258
NameNeurIPS Proceedings

Conference

Conference39th Conference on Neural Information Processing Systems, NeurIPS 2025
Abbreviated titleNeurIPS 2025
Country/TerritoryUnited States
CitySan Diego
Period2/12/25 → 7/12/25
Internet address

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

  1. SDG 3 - Good Health and Well-being
    SDG 3 Good Health and Well-being

Fingerprint

Dive into the research topics of 'Multimodal Disease Progression Modeling via Spatiotemporal Disentanglement and Multiscale Alignment'. Together they form a unique fingerprint.

Cite this