CAMDA: contrastive and meta-path aware learning for microRNA-disease association prediction

The identification of microRNA-disease associations (MDAs) is fundamental to elucidating complex disease pathophysiology. Computational methods for MDA prediction often face challenges in integrating diverse semantic relationships and learning robust representations from sparse data. To address these issues, we propose CAMDA, a framework that learns node representations from multiple meta-path-defined semantic views and the original microRNA–disease association graph. CAMDA utilizes a tri-perspective graph neural architecture with three parallel encoders: a specific encoder for path-dependent semantics, a consensus encoder for view-invariant patterns, and a heterogeneous graph encoder that performs cross-type message passing on the microRNA–disease association graph to capture structural context. A semantic-level attention mechanism aggregates information across meta-path views in the semantic branches, while a multi-view contrastive objective (InfoNCE) encourages cross-view consistency among the three perspectives. The final representations are generated by concatenating embeddings from all three perspectives for downstream prediction tasks. Experiments on the HMDD v3.2 dataset demonstrate that CAMDA achieves an average AUC of 95.81% under five-fold cross-validation. Case studies on three gastrointestinal cancers (esophageal, gastric, and colorectal neoplasms) confirm the model’s ability to identify biologically relevant associations, with validation in established databases supporting the predicted novel microRNA–disease associations.