DBGL: Decay-aware Bipartite Graph Learning for Irregular Medical Time Series Classification

2026-04-12

AI Summary

This study addresses the challenges posed by irregular medical time-series data—specifically heterogeneous sampling rates, asynchronous observations, and variable time intervals—which existing methods struggle to model effectively due to their inability to capture temporal irregularity and variable-specific decay dynamics. To overcome these limitations, this work proposes the first bipartite graph learning framework that jointly models sampling irregularity and variable decay irregularity. By constructing a patient–variable bipartite graph that preserves the original sampling structure and incorporating a node-specific temporal decay encoding mechanism, the model adaptively learns individualized decay rates for each variable over varying time intervals. This approach eliminates the need for manual alignment and more faithfully represents temporal dynamics. Extensive experiments on four public medical time-series datasets demonstrate that the proposed model significantly outperforms current baselines, confirming its effectiveness and robustness.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Reasoning under Uncertainty: Graphical Models
• Data Mining & Knowledge Management: Mining of Spatial, Temporal or Spatio-Temporal Data

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics

Abstract

Irregular Medical Time Series play a critical role in the clinical domain to better understand the patient's condition. However, inherent irregularity arising from heterogeneous sampling rates, asynchronous observations, and variable gaps poses key challenges for reliable modeling. Existing methods often distort temporal sampling irregularity and missingness patterns while failing to capture variable decay irregularity, resulting in suboptimal representations. To address these limitations, we introduce DBGL, Decay-Aware Bipartite Graph Learning for Irregular Medical Time Series. DBGL first introduces a patient-variable bipartite graph that simultaneously captures irregular sampling patterns without artificial alignment and adaptively models variable relationships for temporal sampling irregularity modeling, enhancing representation learning. To model variable decay irregularity, DBGL designs a novel node-specific temporal decay encoding mechanism that captures each variable's decay rates based on sampling interval, yielding a more accurate and faithful representation of irregular temporal dynamics. We evaluate the performance of DBGL on four publicly available datasets, and the results show that DBGL outperforms all baselines.

Problem

• Irregular Medical Time Series
• Temporal Sampling Irregularity
• Variable Decay Irregularity
• Missingness Patterns
• Representation Learning

Innovation

• bipartite graph learning
• temporal decay modeling
• irregular time series
• medical time series classification
• node-specific encoding