In recent years, circRNAs have been found to be closely related to a variety of human diseases. In-depth exploration of their potential associations with diseases is of great value for understanding disease mechanisms and auxiliary diagnosis. To address the association between circRNAs and human diseases, this paper proposes a multi-view contrastive learning model MCHG that integrates attention mechanism and hypergraph structure modeling. The model first constructs a multi-view hypergraph structure based on the circRNA-disease heterogeneous network to capture high-order relationships between nodes. Subsequently, a hypergraph convolutional network is used to extract structured features from different perspectives, and an improved contrastive learning strategy is used to enhance the discriminative ability of feature representation. Furthermore, the model improves the convolutional block attention model, weighting features from two dimensions: channel and space, highlighting key information and suppressing redundant interference. The fused multi-view representations are input into the neural network projection module to achieve efficient prediction of potential associations by reconstructing the circRNA-disease association matrix. Validation experiments on the CircR2Disease dataset show that MCHG achieves an AUC of 0.9394 and an AUPR of 0.9602. Compared with previous models, MCHG can simultaneously model the high-order structural relationships and multi-perspective complementary information in the circRNA-disease heterogeneous network. Furthermore, it improves the quality of feature representation through improved contrastive learning and a two-dimensional attention weighting mechanism, thereby enhancing the accuracy and robustness of potential association prediction. In addition, among the top 25 unknown CDAs predicted by MCHG, 23 were confirmed by relevant literature, indicating that MCHG can effectively predict potential CDAs and provide assistance for further biological wet experiments.
Circular RNAs (circRNAs) participate in gene regulation and disease progression, but experimental mapping of circRNA-disease associations (CDAs) remains costly and incomplete. We developed NARVGA, a hybrid prediction framework that combines non-negative matrix factorisation (NMF) with an adversarially regularised variational graph autoencoder (ARVGA). Functional, semantic and Gaussian interaction-profile kernel similarities are integrated; K-means-derived co-membership graphs reduce diffuse similarity connections; ARVGA learns nonlinear topological embeddings; and NMF captures complementary low-rank association patterns. An Extra Trees classifier then scores candidate circRNA-disease pairs. In the original transductive stratified five-fold benchmark on CircR2Disease, NARVGA achieved an area under the receiver operating characteristic curve (AUC) of 0.9868, an area under the precision–recall curve (AUPR) of 0.9887, 94.32% accuracy and a 94.44% F1-score. Ablation analysis identified cluster sparsification as the largest individual contributor, with additional gains from adversarial regularisation and low-rank augmentation. Without task-specific retuning, the model obtained AUCs of 0.9503 on LncRNADisease and 0.9700 on HMDDv4. Performance was stable across GCN depths and cluster settings, and 19 of the 20 highest-ranked hepatocellular carcinoma candidates had supporting published evidence. NARVGA provides a within-network prioritisation tool for RNA-disease association studies, although prospective validation, hard-negative assessment and independent-cohort testing remain necessary.