Research On The Elimination Of Data Masking In The Detection Of Adverse Drug Reaction Signals In China

After the drug is marketed,Adverse Drug Reaction(ADR)monitoring is the main means of drug re-evaluation,and the Spontaneous Reporting System(SRS)is the main method of ADR monitoring in the world.Countries often use a disproportionate assay to perform signal detection on SRS to find a combination of adverse drug reactions.However,the disproportionate measurement method has certain defects such as data occlusion.This phenomenon is usually caused by the large number of ADR combinations reported in the SRS,and the other ADR combination expectations are too high,and the disproportionate indicators are degraded,resulting in some valuable ADR combinations becoming false.Negative signals,delay signal generation,affect the safety of medication.Therefore,how to eliminate data obscuration and timely and accurate discovery of valuable ADR signals is of great significance to human drug safety.The international mainstream method for the elimination of data obscuration is the removal method,which mainly includes the traditional removal method and the Lasso Logistic Regression(LLR).The traditional removal is mainly to remove the ADR combination of the top 10 reported frequency in the SRS and reuse the disproportionate detection signal to weaken the influence of over-reporting.The removal method has achieved better shielding elimination effect in foreign countries,but the application of these methods in China's ADR data is lacking.In addition,the data quality difference between China's SRS and foreign countries is quite different.Therefore,the characteristics of China's database are proposed to be suitable for China's mask elimination.The method is necessary.The main work of this research is as follows:(1)The most representative traditional removal method for data mask elimination in the world is selected and combined with the Chinese database for empirical research.(2)A new removal strategy is proposed for the characteristics of China's database.Based on the LLR-traditional removal model and the LLR-new removal model,three new methods of mask elimination are used for empirical research.The above work selects the Chinese traditional medicine data of SRS in China as the research object,and compares and analyzes the known signal database to explore the changes of various performance indicators.The research results show that the three innovative methods proposed in this paper have improved F index compared with the traditional removal method,which can effectively improve the signal detection efficiency and effectively reduce the impact of the shadowing effect.Therefore,this article can provide three new ideas and methods for drug testing institutions.