A Hybrid Approach for Drug Interaction Prediction using Knowledge Graphs and Gradient Classifiers

In recent times, the usage of multiple drugs simultaneously to treat complex diseases such as cancer has become common. However, combining drugs that react when taken together may lead to adverse side effects. In our research paper, we introduce a novel machine learning method named AttentiontBoost (AB-DDI), which combines Graph Attention Neural Networks (GATs) and Gradient Boosting Machines (GBMs) to predict drug interactions. We use Simplified Molecular-Input Line-Entry System (SMILES) strings as features for the attention network model and apply a gradient boost classifier to enhance the results. Our proposed model, AB-DDI, gives state-of-the-art results in several metrics.