Bridging Graph and State-Space Modeling for Intensive Care Unit Length of Stay Prediction

2025-08-24

AI Summary

Accurate prediction of length of stay (LOS) for ICU patients is critical for healthcare resource optimization, yet remains challenging due to the heterogeneity and irregular sampling of electronic health records (EHRs). To address this, we propose S²G-Net—a novel dual-path architecture that synergistically models temporal dynamics and topological relationships. Its temporal path employs the Mamba state-space model to capture long-range dependencies in dynamic clinical sequences, while its structural path leverages GraphGPS to integrate multi-source heterogeneous patient similarity graphs—constructed from diagnostic, administrative, and semantic features. Crucially, the graph construction follows principled design principles. Evaluated on MIMIC-IV, S²G-Net achieves state-of-the-art performance, significantly outperforming BiLSTM, Transformer, conventional GNNs, and leading hybrid models across key metrics including MAE and RMSE. Results demonstrate superior effectiveness, robustness to data sparsity and noise, and scalability to large-scale EHR datasets.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

Predicting a patient's length of stay (LOS) in the intensive care unit (ICU) is a critical task for hospital resource management, yet remains challenging due to the heterogeneous and irregularly sampled nature of electronic health records (EHRs). In this work, we propose S$^2$G-Net, a novel neural architecture that unifies state-space sequence modeling with multi-view Graph Neural Networks (GNNs) for ICU LOS prediction. The temporal path employs Mamba state-space models (SSMs) to capture patient trajectories, while the graph path leverages an optimized GraphGPS backbone, designed to integrate heterogeneous patient similarity graphs derived from diagnostic, administrative, and semantic features. Experiments on the large-scale MIMIC-IV cohort dataset show that S$^2$G-Net consistently outperforms sequence models (BiLSTM, Mamba, Transformer), graph models (classic GNNs, GraphGPS), and hybrid approaches across all primary metrics. Extensive ablation studies and interpretability analyses highlight the complementary contributions of each component of our architecture and underscore the importance of principled graph construction. These results demonstrate that S$^2$G-Net provides an effective and scalable solution for ICU LOS prediction with multi-modal clinical data.

Problem

• Predicting ICU length of stay from heterogeneous EHR data
• Integrating state-space models with graph neural networks
• Handling irregularly sampled multi-modal clinical data

Innovation

• Combining state-space models with graph networks
• Using Mamba SSMs for patient trajectory modeling
• Integrating heterogeneous graphs via optimized GraphGPS backbone