Skip to main content

Table 2 Data used for knowledge graph embedding and adverse drug reaction classifier training and testing

From: Prediction of adverse drug reactions based on knowledge graph embedding

Triple Knowledge graph embedding Classifier training Classifier testing Total
(drug, has target, target) 13851 0 0 13851
(drug, has indication, indication) 12238 12238*10\(^\Delta\) 1359 13597
(drug, has side effect, side effect) 125432 125432 1359 126791
  1. \(^\Delta\) To avoid sample imbalance, the drug-indication pairs used for the training of the ADR classification model were replicated and expanded 10 times