Mining And Analysis Of Comorbidity Relationships Based On FAERS Database

With the large-scale use of electronic medical records,a large number of data analysis studies are based on the data provided by electronic medical records.Data studies show that there are many and widespread comorbidities in diseases.The emergence and research of comorbidity play an important role in the mechanism of disease and the prevention and treatment of common diseases in the future.In this paper,we use more advantageous algorithms to dandle abundant data sources,aiming at getting more extensive or difficult common disease pairs.This topic realizes the mining and analysis of the medical record data provided by the Food and Drug Administration Adverse Drug Reaction Reporting System(FAERS).On this basis,we can make the comorbidity pair network and realize visualization.Some of these diseases were classified by cluster analysis.Through early excavation,data analysis and prediction of comorbidity pairs can be realized.All data from the fourth quarter of 2012 to the first quarter of 2017 are selected from FAERS database for processing.Observational Health Data Sciences and Informatics(OHDSI)is selected to clean the data.In the aspect of data processing,it mainly completes data normalization,data extraction,data duplication and so on.After data preprocessing,the problem that the data specific to FAERS database is not clear enough is solved.Previous studies on comorbidity have two problems.First,because of the singularity of data sources,most researchers will do data analysis based on only a single data source;second,it is the shortcomings of data analysis methods,because predecessors mostly use statistical methods,and only handle a pair of data.In this topic,the use of FAERS database resources,the breadth and complexity of data is stronger than previous research.In the choice of methods,the model of association rule algorithm + network graph + data analysis is used to make the conclusion more convincing and more comprehensive for multiple comorbidities.The experimental results show that the network graph analysis results obtained in this subject are valid compared with the known data,and effective results are obtained in data analysis.At the same time,through the experimental study,the network proposed in this paper achieves the visualization effect,which has a certain role in the actual medical treatment.The subject has practical significance.