Graph-Based Modeling of Alzheimer's Protein Interactions via Spiking Neural, Hyperdimensional Encoding, and Scalable Ray-Based Learning

Abstract

This study introduces a novel computational framework for predicting protein-protein interactions (PPIs) in Alzheimers disease by integrating biologically inspired and graph-based learning paradigms. The pipeline combines genetic algorithm-driven feature selection, hyperdimensional encoding for robust semantic representation, and spiking neural networks to capture temporal dynamics. These representations are fused with graph neural network embeddings and processed via scalable nearest-neighbor inference. Unlike conventional models, this multi-view architecture enables biologically grounded prediction under data sparsity, offering interpretable insights into disease-relevant molecular interactions.