Research On Early Warning Of Drug Safety Unexpected Events Under Big Data Environment

Due to the characteristics of "treatment and disease-causing",many drugs are difficult to find out the potential dangers in early clinical trials because of the small test samples,limited observation time and limited scope.In recent years,drug safety unexpected events have occurred frequently,causing serious public health safety problems.According to the statistics of WHO,the deaths due to medication misadventure ranked No.4 to No.6 in the cause of death and showed an upward trend year by year.The incidence and seriousness of drug safety emergencies become more and more serious,which has become the focus of universal attention all over the world.Establishing a scientific and reasonable early warning mechanism has become an issue to be addressed urgently.Under the environment of big data,making full use of all kinds of related information resources to construct the early warning decision knowledge base can provide better decision-making support for the rapid response of emergencies.According to the features of data heterogeneous and multi-source under big data environment,the data resources of drug emergency were collected based on the core of "Drug-Events-Emergency".Organizing and standardizing the data to constructing the data warehouse.Ontology technology and Jena-based rule reasoning mechanism were introduced to build the ontology base which can realize the semantic association and deduce the hidden knowledge in the ontology by using the own rules and self-defined rules of Jena inference.Based on this,constructing an early-warning decision-making knowledge base model for drug emergencies based on the model of "Fact-Concept-Rule".In view of the shortcomings of the traditional early-warning model,three new models for big data environment were put forward which are The risk warning model of single drug based on association analysis,The risk evaluation model of drug category based on clustering analysis and The risk prediction model of new drug based on classification method integrated with the drug safety knowledge base above.Based on the above results,constructing a prototype system for early-warning of drug safety unexpected events by using MySql,Protégé and Java to verify the feasibility and effectiveness.Extracting the characteristics from unexpected events which monitored by public opinion monitoring system.The semantic search module can be used to query the ontology database and output the data.The output result put into the early-warning module to generate signal value and then do the semantic recommendation according to the value to provide decision-making plan for the relevant personnel.