ABSTRACT
In recent years, the safety problems and events of traditional Chinese medicine represented by liver injury have occurred frequently. In particular, Polygonum multiflorum has been widely used and considered as a "non-toxic" tonic traditional Chinese medicine for thousands of years. However, in recent years, frequent reports of liver injury events have attracted widespread attention at home and abroad, which has made unfavorable impacts on traditional Chinese medicine and its international development. Some scho-lars have found that susceptible genes of P. multiflorum on liver injury lay a scientific foundation for formulating rational comprehensive prevention and control measures for liver injury risk of P. multiflorum and its relevant preparations. But what are the risk signals of adverse reactions of P. multiflorum in clinical application? Spontaneous reporting system is an important way to monitor and find adverse drug reaction(ADR) signals after the drug is launched in the market. It can find the ADR signals in time and effectively, and then effectively prevent and avoid the occurrence of adverse drug events. At present, the data mining technique has gradually become the main method of ADR/adverse event(AE) report analysis and evaluation at home and abroad. Specifically, Bayesian confidence propagation neural network in Bayesian method is a commonly used risk signal early warning analysis method. In this paper, BCPNN method was used to excavate the risk signals of adverse reactions of Xinyuan Capsules, a traditional Chinese medicine preparation containing P. multiflorum, such as nausea, diarrhea, rash, dizziness, vomiting, abdominal pain, headache, liver cell damage, so as to provide evidence-based evidence for clinical safe and rational use of drugs.