Explainable Graph Neural Networks for Atherosclerotic Cardiovascular Disease.

Area of Science:

• Cardiovascular Medicine
• Artificial Intelligence
• Machine Learning

Background:

• Atherosclerotic cardiovascular disease (ASCVD) progression and treatment understanding is crucial for clinical decision support systems.
• Trust in these systems requires explainable machine learning models for clinicians, developers, and researchers.
• Graph Neural Networks (GNNs) are emerging for analyzing longitudinal clinical data.

Purpose of the Study:

• To utilize GNNs for modeling and predicting atherosclerotic cardiovascular disease progression.
• To explore explainable AI (XAI) methods for GNNs in this context.
• To investigate the explainability of GNN models for low-density lipoprotein cholesterol levels in long-term ASCVD.

Main Methods:

• Employing Graph Neural Networks (GNNs) to analyze longitudinal clinical trajectories.
• Developing and applying explainable AI (XAI) techniques to GNN models.
• Focusing on modeling and prediction of low-density lipoprotein cholesterol levels.

Main Results:

• Initial project stages focused on methodology development.
• Exploration of GNNs for modeling ASCVD progression.
• Investigation into explainability of GNNs for cholesterol level prediction.

Conclusions:

• GNNs offer a promising approach for modeling complex ASCVD progression.
• Explainable AI (XAI) is essential for clinical adoption of GNN-based decision support.
• Further research is needed to validate and refine these methods for clinical application.