Research On The Key Technologies Of Adverse Drug Reaction Signal Mining In Drug-Drug Interaction

Adverse Drug Reaction(ADR)triggers the problem of drug safety.The occurrence of drug safety events brings additional burden to patients and even affects the safety of patients.A series of ADRs in combined medication account for a large proportion of drug safety problems.Before drugs were put on the market,conventional clinical experimental studies investigating pharmacokinetics were usually used to investigate possible adverse reactions of Drug-Drug Interaction(DDI).However,due to the large number of drug combinations and the limitations of various practical conditions,many adverse reactions could not be discovered,which led to an increase in the ADR risk of combination therapy in the clinical use.Countries all over the world have established their own ADR spontaneous reporting system(SRS),which lays a data foundation for DDI signal mining.However,current studies based on SRS are mainly aimed at the adverse reactions of single medication,while the signal detection of DDI has not enough.Therefore,it is an important task to study of DDI signal detection.ADR monitoring reports of the China Food and Drug Administration(CFDA)from 2010 to 2011 were selected as the research data.This paper took antipsychotic drugs as the research object.The WHO-ART standard term set was used to clean the data,and then the data was conducted by preprocessing,such as extraction,split,eliminate redundancies and so on.As a result,an ADR database of antipsychotic medicines was constructed as the overall sample.Meanwhile,a known DDI-ADR database,including DDI adverse drug reaction data retrieved through the network(www.Drugs.com),was constructed as an objective criterion for performance comparison of each method studied in this paper.First,this paper put forward the false association problem in the data set of combined drug use in the process of data processing.Because ADR reports were fully connected at the time of data splitting,some false drug-adverse reaction combination associations were included in the original data.In order to make the data more realistic,based on the drug-adverse reaction correlation degree in the report of the combined drug use calculation,a false association screening model based on the imbalance theory was established to delete the false association data of these drug-adverse reaction combinations in the original data set.Secondly,this paper conducted DDI signal detection by using the traditional baseline model: the addition model and the multiplication model were applied to the SRS data of our country to mine the DDI signals "drugs A+drugs B?adverse reaction C".Some evaluation indexes based on known DDI-ADR database were established,and the performance of the two baseline models was compared and analyzed.Finally,a MHRA+ method for DDI signal detection was proposed based on MHRA,in which several drugs in combination therapy reports were considered as a new drug.The original MHRA method was used to determine the signal threshold of MHRA+ by screening the multiple values of the signal detection.In this paper,the ADRs reports of antipsychotic medicines in the SRS of China were selected for this study.A screening model of false association among drug-adverse reaction combinations was proposed and used to clean the original data.DDI signal detection was conducted by using the addition model and multiplication model of the baseline model and the improved MHRA+ method.The performance of each method was compared and analyzed.The experimental results show that compared with the baseline model,the MHRA+ method proposed has a higher F index,which provides a reference method for ADRs signal mining in Drug-Drug Interaction.