miRNA-disease association

The accurate identification of miRNA-disease associations plays a crucial role in biomedical research and clinical applications. However, most research focuses on the existence of the association, without conducting further exploration. In this study, we propose a novel statistical meta-path contrastive learning-based approach (SMCLMDA), which aims to accurately identify the multidimensional relationships(up/down-regulation and causal/non-causal) between miRNAs and diseases. SMCLMDA first obtains the local structural information of miRNAs and diseases based on Node2Vec and uses it as the initial node input of the graph convolutional neural network (GCN). Then, the meta-path view constructed by the statistical method further enhances the representation of the similarity view via a contrastive learning strategy. Finally, SMCLMDA calculates the predicted probability of the multidimensional relationships of miRNA-disease via a multilayer perceptron(MLP). The experimental results demonstrate that SMCLMDA surpasses current state-of-theart computational methods in predicting traditional associations, as well as up/down-regulation and causal/non-causal relationships. The case study additionally validates the effectiveness of SMCLMDA in identifying potential miRNA-disease
associations.