Prediction of Therapeutic Usage of Jamu Based on the Composition of Metabolites using Convolutional Neural Network

Herbal medicine has contributed to the development of modern medicine by examining the chemical content of medicinal plants. Jamu is a traditional herbal medicine from Indonesia that has been consumed by people for generations to maintain health and cure diseases. Its prescription is composed of various plants, and each plant consists of various metabolites. A metabolite is a small molecule and very useful in the process of drug discovery. In this study, we predict the therapeutic usage of Jamu based on the composition of metabolites from plants that are used as Jamu ingredients. Then, we identify potential metabolites for each therapeutic usage based on the best prediction model generated by the Convolutional Neural Network technique. In conclusion, the average accuracy resulted from the prediction model of Convolutional Neural Network is 92.48% using 10-fold cross-validation. Additionally, we extracted 32 potential metabolites from eight disease classes, and some of them have been validated by published literature.