Risk Grading Prediction And Modeling Analysis Of National Stroke Screening Data

Aiming at the problems existing in the national stroke screening data,such as large amount of data not graded,low data utilization and complex maintenance,Based on the related classification algorithm and data warehouse technology,by accurately predicting the risk level of screening data and establishing the dimensional model of data warehouse,accurate statistical analysis and efficient utilization of stroke screening data can be realized,and data support can be provided for clinical research and medical decision-making of stroke prevention and treatment.In order to solve the problem that the large screening data is ungraded due to the shortage of medical resources,taking the national stroke history screening data as the research object,and a C5.0 algorithm and BP neural network are proposed,a dangerous level prediction model combining C5.0 algorithm and BP neural network algorithm is proposed.Firstly,the initial sample data is obtained by stratified sampling and oversampling,and then the C5.0 algorithm with feature.classification is used to select the characteristic variables which with strong correlation,and use this as the initial value of the training of neural network model,using hypertension and diabetes,the neural network with 8 main related characteristic variables such as hypertension and diabetes was trained and classified and predicted,which effectively improved the accuracy of hazard classification under multivariate and missing information data.The test results show that the overall classification accuracy of the C5.0-BP combined prediction model is 93.68%,which is 8.65%higher than the current China Stroke Data Center based on the C4.5 algorithm classification model.The comprehensive evaluation of the low-risk and high-risk level prediction.The indicators increased by 37.5%and 63%respectively.The model achieves a more accurate risk grading prediction and promotes the prevention and intervention of stroke diseases.In view of the shortcomings of the current national stroke prevention data platform,such as low interaction and complex maintenance,according to the risk grading prediction data,the merging,cleaning and transformation rules were designed,and the data warehouse dimension model including 4 analysis topics and 50 analysis dimensions was constructed,and the analysis results were displayed by using BI service.From the analysis results,the screening data warehouse dimension model realizes the analysis and display of different themes and dimensions,and with has high interactivity.At present,the dimensional model has been partially applied in stroke data analysis of the National Health Commission of China,providing data support for national stroke screening prevention and treatment.At the same time,it provides data support for Chia's stroke screening prevention and treatment work,and provides data analysis services for management decision-making and stroke epidemiological research.