Advances in the Application of Machine Learning and Deep Learning in Drug–Drug Interaction Prediction
TL;DR
The current application status of artificial intelligence in DDI prediction can promote existing pharmacological theories and provide practical references for clinical medication, which can promote existing pharmacological theories and provide practical references for clinical medication.
Abstract
. While drug combination therapy can enhance efficacy, it may also trigger drug interactions. Traditional detection methods are time-consuming and costly, making artificial intelligence an important auxiliary technology. This article reviews traditional machine learning methods and deep learning methods perspectives. Traditional machine learning methods can use the Drug – Drug Interaction (DDI) information database of pharmacokinetics (PK) for prediction, but this method has a relatively high bias risk for all models. Deep learning methods often establish models based on neural networks, and the results are relatively more accurate. Artificial intelligence has played an important role in DDI prediction, but there are still limitations such as insufficient consideration of clinical factors and inconsistent use of data sets, which are expected to be resolved. This article summarizes the current application status of artificial intelligence in DDI prediction, which can promote existing pharmacological theories and provide practical references for clinical medication.