Beyond Feature Importance: Feature Interactions in Predicting Post-Stroke Rigidity with Graph Explainable AI

2025-04-10

AI Summary

This study addresses the challenges of early prediction and poor interpretability in post-stroke muscle rigidity. We propose an interpretable AI framework based on graph neural networks (GNNs). Specifically, we construct a clinical feature interaction graph using the HCUP 519K dataset and pioneer the integration of Graphormer and graph attention networks (GATs), augmented with rigorous interpretability analysis. This approach uncovers synergistic interactions among key clinical features—such as the NIH Stroke Scale (NIHSS) and APR-DRG mortality risk score—revealing pathophysiological mechanisms underlying rigidity onset. Compared to baseline models—including logistic regression, XGBoost, and Transformer—the optimal GNN achieves an AUROC of 0.75 (+0.08–0.12 improvement), demonstrating statistically significant gains in predictive performance. Moreover, our framework identifies clinically meaningful predictive factors and their interaction pathways, enabling transparent, evidence-based decision support for early intervention and personalized rehabilitation planning.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Reasoning under Uncertainty: Graphical Models

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

This study addresses the challenge of predicting post-stroke rigidity by emphasizing feature interactions through graph-based explainable AI. Post-stroke rigidity, characterized by increased muscle tone and stiffness, significantly affects survivors' mobility and quality of life. Despite its prevalence, early prediction remains limited, delaying intervention. We analyze 519K stroke hospitalization records from the Healthcare Cost and Utilization Project dataset, where 43% of patients exhibited rigidity. We compare traditional approaches such as Logistic Regression, XGBoost, and Transformer with graph-based models like Graphormer and Graph Attention Network. These graph models inherently capture feature interactions and incorporate intrinsic or post-hoc explainability. Our results show that graph-based methods outperform others (AUROC 0.75), identifying key predictors such as NIH Stroke Scale and APR-DRG mortality risk scores. They also uncover interactions missed by conventional models. This research provides a novel application of graph-based XAI in stroke prognosis, with potential to guide early identification and personalized rehabilitation strategies.

Problem

• Predicting post-stroke rigidity using feature interactions
• Comparing graph-based models with traditional prediction methods
• Identifying key predictors for early stroke intervention

Innovation

• Graph-based explainable AI for feature interactions
• Graphormer and Graph Attention Network models
• Identifies key predictors and feature interactions