Research Of Cohorts And Therapies Based On Electronic Health Record Data

With the development of medical informatization,more and more people focus on the research of electronic medical record data.It is hoped that the disease patterns found in the electronic medical record data can help to diagnose and treat diseases.Breast cancer is one of the most common malignant tumors.Therefore,analyzing the clinical data of breast cancer and discovering the regularity have great significance for the treatment and prognosis of patients.Based on similar patient groups and treatment methods,system is built by using the corresponding visual analysis methods and the prediction model.In order to explore the patterns in the cohort,this paper designs the similar patient cohorts' visual analysis,and explores the correlation features by comparing the cohorts.The users can select the features of the patient examination report,the appropriate dimension reduction algorithm and clustering algorithm according to the scatter plot,and then analyze the different groups by the visualization methods such as Nightingale graph,word cloud and timeline.Through the specific case analysis,this paper verifies the method can help users find associations in the data.In order to design a treatment plan,this paper defines the visual analysis of attribute relevance,which can explore the attribute relevance and get treatment plan based on this,can assist users in preoperative decision-making.Parallel coordinates allow the users to explore attribute dependencies independently.The matrix thermogram shows the feature weights trained by the predictive model.The predictive model uses the SVM model,and the predicted results are displayed through the classification map.Finally,by analyzing the attribute relevance and exploring the outcome of the treatment plan,this system can assist the users in preoperative decision-making.