Community Graph Convolution Neural Network for Alzheimer's Disease Classification and Pathogenetic Factors...

Area of Science:

• Neuroscience
• Computational Biology
• Artificial Intelligence

Background:

• The brain functions as a complex neural network where regions and genes collaborate for information processing.
• Understanding these intricate collaborations is crucial for neurological disorder research, particularly Alzheimer's disease (AD).

Purpose of the Study:

• To develop a novel deep learning framework for analyzing brain region gene community networks (BG-CN).
• To investigate information transmission mechanisms within and between these communities.
• To improve the diagnosis and identification of causal factors for Alzheimer's disease (AD).

Main Methods:

• Abstracted collaboration correlations into a brain region gene community network (BG-CN).
• Developed an affinity aggregation model to describe intercommunity and intracommunity information transmission.
• Designed the community graph convolutional neural network (Com-GCN) with specialized convolution operations.

Main Results:

• The Com-GCN model demonstrated improved interpretability and classification performance on the AD neuroimaging initiative (ADNI) dataset.
• The Com-GCN architecture aligns better with physiological mechanisms of information transmission.
• Identified potential lesioned brain regions and disease-causing genes associated with AD.

Conclusions:

• Com-GCN offers a powerful tool for analyzing complex brain networks and information flow.
• The findings can aid in precision medicine and drug design for AD.
• This approach serves as a valuable reference for studying other neurological disorders.