Data Mining Of Electronic Medical Records Based On Association Rules-Diabetes And Its Complications As An Example

[Purpose] With the rapid development of computer technology,data storage technology,network technology and the deepening application in the medical field,the hospital informatization is developing deeply,which producted massive clinical data,in which the data in the electronic medical record system is particularly rich.At the same time,the medical system is generally faced with the problem that “more information,less knowledge”,how to use advanced techniques and methods to discover the information behind the massive data,how to learn from these data to improve healthcare quality is a current research focus.Using diabetes as an example,this study applied association rules into the field of medicine,in order to discover the associations between diabetes and its complications,improve clinical quality and efficiency,and provide a reference for researches about other diseases.[Methods] Data was collected from the electronic medical record system of a hospital in Qingdao.A total of 9094 patients' records with diabetes inpatient treatment records were collected.Firstly,MySQL relational database technology was used to process the original data for data cleansing,data integration,discretization and other operations to ensure the data validity and integrity to be analyzed.Then the Apriori correlation algorithm was applied to analyze the relationship of diabetes and its complications on the data on the Xcode platform after seting the minimum confidence and minimum support.[Results] Through the analysis of Apriori correlation,we get the frequent item-sets,including six 1-item-set,nine 2-item-sets and four 3-item-sets.By limiting the minimum support degree and the minimum confidence degree,we get 13 correlations rules about diabetes and its complications And there are five strong association rules beyond the minimum support of 20%.[Conclusions] According to the association analysis of this study,diabetes has strong correlation with coronary heart disease,hypertension,fatty liver,chronicarterial occlusion and abnormal lipid metabolism.Through the analysis of the preceding and subsequent items in the association rule table,diabetes have great potential to lead to high blood pressure,coronary heart disease,lipid metabolism,fatty liver,and a variety of complications between the existence of a more closely related relationship.Among them,"coronary heart disease,fatty liver,high blood pressure,lipid metabolism abnormalities" this four conclusions cater to the current medical research results,"chronic arterial occlusive disease" is the newly discovered diabetic complications.