Prediction of Alzheimer’s Micro Rna – Disease Causal Association Using Graph Neural Networks

Introduction: Alzheimer's disease (AD) is a neurodegenerative disorder characterized by the presence of tau protein tangles and beta-amyloid plaques, which disrupt brain function and lead to neuronal loss. Objective: To elucidate the molecular mechanisms underlying AD and identify novel therapeutic targets, focusing on establishing causal links between microRNAs and AD, critical for early detection and disease monitoring. Methods: The study utilized the HMDD V4.0 database to examine a dataset concerning causality in microRNA-disease associations, comprising microRNA expression profiles and clinical information from AD patients and healthy individuals. The dataset was subjected to graph neural networks to predict causal interactions between microRNAs and AD. Results: The graph neural network model demonstrated robust performance in predicting Alzheimer's disease, achieving a precision of 69.44%, recall of 83.33%, and F1 score of 75.75%, indicating its potential in identifying biomarkers or therapeutic targets using microRNA data. Conclusion: The graph neural network model for predicting causal associations between Alzheimer's disease and microRNAs shows promise. However, further investigation is warranted regarding dataset balance, adjustments to classification thresholds, incorporation of additional features, and exploration of advanced model architectures.