Research On The Association Between Disease And Drug Based On Scientific Literature Mining

Objective:The biomedical entity is a name,terminology or concept that contains genes,diseases,drugs,etc.that appear in scientific literature.It is implicit knowledge in literature that understanding its association is significant for scientific research.However,such knowledge is overwhelmed by the large amount of submerged literature,and there is an urgent need for an effective knowledge management to show them quickly.In view of this,we research on the association between disease and drug based on scientific literature mining.Methods: 1.Literature AnalysisThrough the collection,identification,collation of relevant literature,analysis of the current domestic and foreign related research history,status and problems.In the reading,collation,induction,analysis of these documents on the basis of reference,learn from others research results,thus forming their own research framework.2.Programming Languages and DatabaseUsing the Java and Python programming language and database technology to PubMed ID related to the literature information download,bulk data order,and in the MySQL database were built library.3.Biomedical entity identificationIdentification of disease entities and drug entities using dictionary-based matching methods.4.Information metrology methodUsing the Python self programming language,the coexistence relationship between the disease and the drug entity is based on the co-occurrence relationship in the information metrology.The word frequency analysis and the co-word analysis are used to analyze the association of the disease and the drug entity.5.Social network analysis methodUsing the Pajek which is a social network analysis tool analysis the co-occurrence network using the macro-level and micro-level indicators.Comparative analysis in Micro-level indicators contain Centrality(Degree Centrality,Closeness Centrality,Betweenness Centrality)and so on.Finally,we use Gephi to visualize the co-occurrence network.Results and conclusions:The results show that the biomedical entity recognition and association discovery methods used in this study can help researchers to quickly detect hidden associations from large-scale biomedical texts,and have good generalization,so also apply to find association between diseases and genes,Gene and drug and other biomedical entities.