Explainable Graph-theoretical Machine Learning: with Application to Alzheimer's Disease Prediction

2025-03-20

AI Summary

This study addresses two key challenges in Alzheimer’s disease (AD): the difficulty of detecting early metabolic brain network alterations and the low accuracy of individualized cognitive prediction. We propose XGML, an interpretable graph machine learning framework that constructs subject-specific metabolic brain graphs from FDG-PET data. Functional distances between regions are modeled by integrating kernel density estimation with dynamic time warping (DTW), and a graph neural network enables subgraph-level importance quantification and joint regression over multiple cognitive outcomes. XGML overcomes limitations of conventional group-averaged templates and static thresholding, enabling discovery of individualized biomarkers. Evaluated on the ADNI cohort, XGML significantly predicts eight cognitive scores (e.g., CDRSB: r = 0.74; ADAS11: r = 0.73) and identifies domain-specific critical edges, offering a novel paradigm for elucidating AD network mechanisms and enabling precision early intervention.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Cognitive Modeling & Cognitive Systems: Adaptive Behavior

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Alzheimer's disease (AD) affects 50 million people worldwide and is projected to overwhelm 152 million by 2050. AD is characterized by cognitive decline due partly to disruptions in metabolic brain connectivity. Thus, early and accurate detection of metabolic brain network impairments is crucial for AD management. Chief to identifying such impairments is FDG-PET data. Despite advancements, most graph-based studies using FDG-PET data rely on group-level analysis or thresholding. Yet, group-level analysis can veil individual differences and thresholding may overlook weaker but biologically critical brain connections. Additionally, machine learning-based AD prediction largely focuses on univariate outcomes, such as disease status. Here, we introduce explainable graph-theoretical machine learning (XGML), a framework employing kernel density estimation and dynamic time warping to construct individual metabolic brain graphs that capture the distance between pair-wise brain regions and identify subgraphs most predictive of multivariate AD-related outcomes. Using FDG-PET data from the Alzheimer's Disease Neuroimaging Initiative, XGML builds metabolic brain graphs and uncovers subgraphs predictive of eight AD-related cognitive scores in new subjects. XGML shows robust performance, particularly for predicting scores measuring learning, memory, language, praxis, and orientation, such as CDRSB ($r = 0.74$), ADAS11 ($r = 0.73$), and ADAS13 ($r = 0.71$). Moreover, XGML unveils key edges jointly but differentially predictive of several AD-related outcomes; they may serve as potential network biomarkers for assessing overall cognitive decline. Together, we show the promise of graph-theoretical machine learning in biomarker discovery and disease prediction and its potential to improve our understanding of network neural mechanisms underlying AD.

Problem

• Early detection of metabolic brain network impairments in Alzheimer's disease.
• Overcoming limitations of group-level analysis and thresholding in FDG-PET data.
• Predicting multivariate AD-related outcomes using explainable graph-theoretical machine learning.

Innovation

• Uses kernel density estimation for graph construction
• Applies dynamic time warping to capture brain region distances
• Identifies predictive subgraphs for Alzheimer's disease outcomes