Saved in:
Bibliographic Details
Main Authors: Lyu, He, Zeng, Huolin, Wang, Junren, Yang, Huazhen, He, Linchao, Chen, Yong, Li, Zhirui, Maier, Andreas, Bayer, Siming, Song, Huan
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.03570
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910191917727744
author Lyu, He
Zeng, Huolin
Wang, Junren
Yang, Huazhen
He, Linchao
Chen, Yong
Li, Zhirui
Maier, Andreas
Bayer, Siming
Song, Huan
author_facet Lyu, He
Zeng, Huolin
Wang, Junren
Yang, Huazhen
He, Linchao
Chen, Yong
Li, Zhirui
Maier, Andreas
Bayer, Siming
Song, Huan
contents Real-world clinical data is inherently multimodal, providing complementary evidence that mirrors the practical necessity of jointly assessing multiple related outcomes. Although multi-task learning can improve efficiency by sharing information across outcomes, existing approaches often fail to balance shared representation learning with outcome-specific modeling. Hard parameter sharing can trigger negative transfer when task gradients conflict, while flexible sharing may still entangle shared and task-specific signals. To address this, we propose a multi-task framework built on a unified Transformer for multimodal fusion, augmented with Orthogonal Task Decomposition (OrthTD) to split patient representations into shared and task-specific subspaces and impose a geometric orthogonality constraint to reduce redundancy and isolate task-specific signals. We evaluated OrthTD on a real-world cohort of 12,430 surgical patients for predicting four outcomes. OrthTD achieved average AUC (area under the receiver operating characteristic curve) of 87.5% and average AUPRC (area under the precision-recall curve) of 37.2%, consistently outperformed advanced tabular and multi-task methods. Notably, OrthTD achieves substantial gains in AUPRC, indicating superior performance in identifying rare events within imbalanced clinical data. These results suggest that enforcing non-redundant shared and task-specific representations can improve multi-outcome prediction from multimodal clinical data.
format Preprint
id arxiv_https___arxiv_org_abs_2605_03570
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Disentangling Shared and Task-Specific Representations from Multi-Modal Clinical Data
Lyu, He
Zeng, Huolin
Wang, Junren
Yang, Huazhen
He, Linchao
Chen, Yong
Li, Zhirui
Maier, Andreas
Bayer, Siming
Song, Huan
Machine Learning
Artificial Intelligence
Real-world clinical data is inherently multimodal, providing complementary evidence that mirrors the practical necessity of jointly assessing multiple related outcomes. Although multi-task learning can improve efficiency by sharing information across outcomes, existing approaches often fail to balance shared representation learning with outcome-specific modeling. Hard parameter sharing can trigger negative transfer when task gradients conflict, while flexible sharing may still entangle shared and task-specific signals. To address this, we propose a multi-task framework built on a unified Transformer for multimodal fusion, augmented with Orthogonal Task Decomposition (OrthTD) to split patient representations into shared and task-specific subspaces and impose a geometric orthogonality constraint to reduce redundancy and isolate task-specific signals. We evaluated OrthTD on a real-world cohort of 12,430 surgical patients for predicting four outcomes. OrthTD achieved average AUC (area under the receiver operating characteristic curve) of 87.5% and average AUPRC (area under the precision-recall curve) of 37.2%, consistently outperformed advanced tabular and multi-task methods. Notably, OrthTD achieves substantial gains in AUPRC, indicating superior performance in identifying rare events within imbalanced clinical data. These results suggest that enforcing non-redundant shared and task-specific representations can improve multi-outcome prediction from multimodal clinical data.
title Disentangling Shared and Task-Specific Representations from Multi-Modal Clinical Data
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2605.03570