Fig. 1From: Multi-type feature fusion based on graph neural network for drug-drug interaction predictionOverview of MFFGNN, where \(\bigoplus\) is sum. The MFFGNN uses SMILES sequences and molecular graphs as inputs to the model, and then extracts the intra-drug features through the MGFEM and SSFEM modules, respectively. Then, MFFGNN fuses the intra-drug features and external DDI features through MFFM module to obtain the updated drug features. Finally, the final predicted value is obtained by DDI predictorBack to article page