Drug to Drug Interaction Classifier

An innovative approach was developed to address a crucial challenge in drug-drug interaction research. While existing state of the art link prediction models rely on prior knowledge of a drug's interaction with other drugs, our solution utilizes the CatBoost to classify potential interactions based solely on intrinsic properties.

We developed a new method for predicting drug interactions using the CatBoost algorithm that relies solely on intrinsic properties, rather than prior knowledge of a drug's interactions. We achieved a high accuracy of 0.85 and an AUC-ROC score of 0.86. This breakthrough provides a more efficient and cost-effective approach to predicting drug interactions, particularly for new drugs without prior interaction data.