Drug-drug interactions can prompt various health issues that take place when a person intakes multiple drugs simultaneously. Predicting these events in advance can save lives. A polynomial kernel SVM was proposed for predicting drug-drug interactions (DDIs) between a drug pair. A unique fingerprint of each drug was considered, and then, the fingerprints of every drug pair were combined into another unique fingerprint using a novel method which then acts as a feature vector for the machine learning model. The SVM implemented gave us an accuracy of 91.6%. Moreover, the proposed model could also predict novel interactions between drugs not present in the dataset.
Paper Link: https://link.springer.com/chapter/10.1007/978-981-19-9379-4_23