Classification modeling of support vector machine (SVM) and random forest in predicting pharmacodynamics interactions

Abstract

Abstract Drug-drug interaction (DDI) is a drug’s effectiveness that can affect the body’s response to the treatment process. DDI occurs when food, drinks, chemicals, and other drugs change the effectiveness of a drug that is given simultaneously. One type of DDI is pharmacodynamics interactions. This interaction is difficult to detect and is very dangerous to humans. Therefore it is necessary to do classification modeling to identify pharmacodynamics interactions based on the value of Side Effect Similarity (SES), Chemical Similarity (CS), and Target Protein Connectedness (TPC). The Support Vector Machine (SVM) and random forest classification method that can be used to predict pharmacodynamics interactions. This study aims to find the best classifications technique by first applying the scaling process, variables interaction, resampling technique, and binarization technique. Best on the analysis result obtained by the random forest is the best model with the highest accuracy and AUC value to other models. The accuracy and AUC values for the best models are 89.93% and 79.96%.