Integrated action crossing method for Drug-Drug Interactions prediction in noncommunicable diseases based on neural networks

Drug-Drug Interactions (DDI) is a cause of treatment inefficacy and toxicity. The most DDI involve drug metabolism which related to enzyme and transporter protein. Drug-enzyme actions that alter the metabolism of other drugs consist of substrate, inhibitor and inducer. Non-communicable diseases (NCDs) are the leading cause of death, drugs that are used in NCDs can increase interaction probability because their long-term usage. This paper proposes Integrated Action Crossing (IAC), a new attribute generation method for DDIs prediction in NCDs. Drugs attributes in NCDs categories were extracted. The actions of enzymes and transporter proteins were crossed for generating dataset for prediction model creation. Neural network (NN) and others machine learning were investigated. Five-fold cross validation was performed for evaluaing the prediction model performance. The results showed that 2 layers NN obtained the best performance of NCDs DDIs prediction model at the accuracy of 83.15%.