🤖 AI Summary
Existing methods for predicting critical illness severity in ICU patients neglect inter-patient relational structures. Method: We propose a dynamic self-constructing graph neural network (GNN) that dynamically builds patient similarity graphs by jointly leveraging feature and structural similarity. A hybrid GNN architecture—integrating GCN, GraphSAGE, and GAT—jointly models local neighborhood patterns and global topological structures; an attention mechanism enables adaptive weighting of clinically relevant features and enhances prediction interpretability. Results: Evaluated on the MIMIC-III dataset (6,000 ICU admissions), our model achieves an AUC-ROC of 0.94, significantly outperforming baseline and single-GNN models. It also improves recall and provides clinically interpretable predictions, establishing a new paradigm for early critical illness warning that balances accuracy and clinical trustworthiness.
📝 Abstract
Accurately predicting the criticalness of ICU patients (such as in-ICU mortality risk) is vital for early intervention in critical care. However, conventional models often treat each patient in isolation and struggle to exploit the relational structure in Electronic Health Records (EHR). We propose a Similarity-Based Self-Construct Graph Model (SBSCGM) that dynamically builds a patient similarity graph from multi-modal EHR data, and a HybridGraphMedGNN architecture that operates on this graph to predict patient mortality and a continuous criticalness score. SBSCGM uses a hybrid similarity measure (combining feature-based and structural similarities) to connect patients with analogous clinical profiles in real-time. The HybridGraphMedGNN integrates Graph Convolutional Network (GCN), GraphSAGE, and Graph Attention Network (GAT) layers to learn robust patient representations, leveraging both local and global graph patterns. In experiments on 6,000 ICU stays from the MIMIC-III dataset, our model achieves state-of-the-art performance (AUC-ROC $0.94$) outperforming baseline classifiers and single-type GNN models. We also demonstrate improved precision/recall and show that the attention mechanism provides interpretable insights into model predictions. Our framework offers a scalable and interpretable solution for critical care risk prediction, with potential to support clinicians in real-world ICU deployment.