arXiv cs.CL· Zihan Liang, Ziwen Pan, Ruoxuan Xiong·· 11 小时前AI 评分24
面向非随机模态缺失的多模态临床记录因果表征学习
Causal Representation Learning from Multimodal Clinical Records under Non-Random Modality Missingness
AI 导读
研究者提出一种因果表征学习框架，利用多模态临床记录中的观测数据与信息性缺失模式，在 MIMIC-IV 和 eICU 上相对最强基线将再入院 AUC 最高提升 4.3%、ICU 入住 AUC 提升 2.7%。该框架包含 MMNAR 感知的模态融合、带潜在对齐的模态重建，以及带校正器的多任务结局预测模型。在 MIMIC-IV 队列中，35.15% 的患者没有合格的出院小结。
正文 · 原文
Abstract:Clinical notes contain rich patient information, such as diagnoses or medications, making them valuable for patient representation learning. Recent advances in large language models have further improved the ability to extract meaningful representations from clinical texts. However, clinical notes are often missing. For example, in our MIMIC-IV cohort, 35.15% of patients have no eligible discharge summaries. In such cases, representations can be learned from other modalities such as structured data, chest X-rays, or radiology reports. Yet the availability of these modalities is influenced by clinical decision-making and varies across patients, potentially producing modality missing-not-at-random (MMNAR) patterns. We propose a causal representation learning framework that leverages observed data and informative missingness in multimodal clinical records. It consists of: (1) an MMNAR-aware modality fusion component that integrates structured data, imaging, and text while conditioning on missingness patterns to capture patient health and clinician-driven assignment; (2) a modality reconstruction component with latent alignment to preserve modality information in the learned representation; and (3) a multitask outcome prediction model with a rectifier that corrects for residual bias from specific modality observation patterns. Comprehensive evaluations across MIMIC-IV and eICU show consistent gains over the strongest baselines, achieving relative AUC improvements of up to 4.3% for hospital readmission and 2.7% for ICU admission.
| Comments: | To appear in Proc. of EMNLP 2025 (21 pages) |
| Subjects: | Machine Learning (cs.LG); Computation and Language (cs.CL); Methodology (stat.ME) |
| Cite as: | arXiv:2509.17228 [cs.LG] |
| (or arXiv:2509.17228v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2509.17228 arXiv-issued DOI via DataCite |
Submission history
From: Zihan Liang [view email]
[v1]
Sun, 21 Sep 2025 20:34:52 UTC (229 KB)
[v2]
Fri, 6 Nov 2026 00:23:13 UTC (330 KB)
来源：arXiv cs.CL · arxiv.org