Integrates variational autoencoders and generative adversarial networks for feature extraction and manifold learning in microbe–drug interaction prediction.

Chemical structure–based graph convolutional model specifically designed for drug–gut microbiota association prediction.

Dual-hypergraph contrastive model with hierarchical attention capturing higher-order structural relationships.

Knowledge graph contrastive learning approach for multi-relational microbe–drug inference with enhanced interpretability.

Graph Transformer-based contrastive learning model leveraging augmented graph views for improved prediction accuracy.

Structure-aware contrastive learning with self-paced negative sampling for robust representation learning.

Hybrid kernel fusion with GCN to integrate multiple similarity sources and improve generalization performance.

Combines local and global graph neural network embeddings for improved representation learning and prediction accuracy.

Graph convolution + attention network hybrid model for microbe–drug association prediction with enhanced performance.

Graph attention and random forest hybrid approach for interpretable microbe–drug association prediction.

Integrates similarity tensor decomposition and network propagation with transformer architecture for enhanced stability.

Proposes a novel model leveraging multi-scale topology and position feature learning combined with relationship-aware graph reasoning to predict drug-related microbes.

Integrates singular value decomposition and Node2vec for comprehensive feature extraction and fusion.

Matrix factorization approach with three-layer heterogeneous network for comprehensive association modeling.

Minimizes matrix nuclear norm with graph attention network for potential microbe–drug association prediction.

Graph attention network + CNN-based classifier for human microbe–drug association prediction.

Joint deep autoencoder with subgraph augmentation for robust representation learning and prediction.

Neighborhood-based inference and restricted Boltzmann machine for microbe–drug association prediction.

Graph attention network + sparse autoencoder for potential microbe–drug association prediction.

Nearest Neighbor Attention Network leveraging bipartite graph networks and attention mechanisms.

Variational graph autoencoder learning latent microbe–drug associations on heterogeneous graphs.

Stacked autoencoder with multi-head attention mechanism for comprehensive feature learning and prediction.

First GCN model for microbe–drug prediction, integrating similarity networks and graph convolution.

Network-based model using KATZ measure for predicting drug–gut microbiota associations.