Fig. 1
From: DeSIDE-DDI: interpretable prediction of drug-drug interactions using drug-induced gene expressions
Overview of the drug-drug interaction (DDI) prediction model. A Datasets are preprocessed for each feature generation model and DDI prediction model. B Given drug-pair and side-effect data, each drug feature is obtained using a feature generation model and a side effect is embedded to latent space; then, drugs pass through neural networks having gated linear units to consider the pair information. Finally, using latent representations of drugs and the side effect, the side-effect score is calculated. C The model is evaluated considering three DDI cases. D, E The independent dataset from DrugBank is used for external validation