BrainNNExplainer: An Interpretable Graph Neural Network Framework for Brain Network based Disease Analysis

The proposed BrainNNExplainer trained in three-steps: the initial training of BrainNN on the original data, the explanation generation based on trained BrainNN, and the further adjustment of BrainNN based on the explanations.
ICML 2021 Workshop on Interpretable Machine Learning in Healthcare

Interpretable brain network models for disease prediction are of great value for the advancementof neuroscience. GNNs are promising to model complicated network data, but they are prone to overfitting and suffer from poor interpretability, which prevents their usage in decision-critical scenarios like healthcare. To bridge this gap, we propose BrainNNExplainer, an interpretable GNN framework for brain network analysis. It is mainly composed of two jointly learned modules: a backbone prediction model that is specifically designed for brain networks and an explanation generator that highlights disease-specificprominent brain network connections. Extensive experimental results with visualizations on two challenging disease prediction datasets demonstrate the unique interpretability and outstanding performance of BrainNNExplainer.