Implicit Association Discovery And Measurement Of Health Medical Data

The rapid growth of the types and sizes of health and medical data in the past decade has made it possible to analyze potential underlying relationships and regulatory mechanisms.It is of great significance for early warning of diseases,assistance in diagnosis and treatment,and personalized medicine.As the State Council and the National Health and Family Planning Commission have successively issued documents on promoting the development of health care informatization in the last three years,China has completely entered the initial stage of large healthcare data,in-depth understanding of health care data has become the primary task of the current stage.At the present stage,the in-depth application of the theories and methods of health-related medical data association mining still face the major theoretical and technical bottlenecks of diverse data types and low mining efficiency.Due to the differences in diagnostics and data collection methods in clinical medicine,the types of health care data are diverse.According to whether standardized,health care data are classified into semantic and non-semantic,and there are including the following challenges:(1)Excavation of interesting association rules for mass health medical data.Semantic health data has a large number of even massive data items and transaction numbers,including diseases,various vital signs,and other information.It is of great significance to search for interesting relationships among them for the prevention and diagnosis of diseases.Traditional data analysis methods face two problems:the difficulty of responding to the scale of health care data;the results often fail to fully reflect the relevance.Therefore,it is of utmost importance to propose appropriate association mining solutions based on semantic health care data to fully and efficiently mine interesting associations between diseases and vital signs.(2)Synchronous Analysis Metric Framework Based on Complex Networks.For non-semantic health medical data,this paper uses the network constructed by mutual information and exponential smoothing rules to represent the complex dynamic system of non-semantic health medical data,and then measures its synchronization through complex network analysis methods,so that a full understanding of non-semantic can be achieved.Synchronous law and internal state evolution of health medical data system.Based on the above issues,this article has carried out research on the discovery and measurement of health care data associations.The specific contents include:(1)Analysis of complex association rules based on FP-Growth.For mass health data with semantic health,this paper designs an efficient association rule based on FP-Growth,which takes into account the negative correlation of data.It makes full use of the high efficiency of FP-Growth algorithm and it is suitable for large data sets.Branch strategy can significantly reduce the impact of increasing complexity of the algorithm after adding a negative correlation,and explore more meaningful relationships while ensuring the efficiency of the algorithm.(2)Synchronization analysis metrics framework based on complex networks.For non-semantic health medical data,this paper uses the network constructed by mutual information and exponential smoothing rules to represent the complex dynamic system of non-semantic health medical data,and then measures its synchronization through complex network analysis methods,so that a full understanding of non-semantic can be achieved.Synchronous law and internal state evolution of health medical data system.The experimental results show that:(1)Compared with the original algorithm,PNFP-Growth algorithm can effectively mine the potential positive and negative correlations in the system,and deal with the performance of large-scale complex data;(2)The synchronization analysis framework based on complex networks can Effectively constructing the brain function network,compared with the traditional time series analysis method,it can obtain the global synchronization,local synchronization and synchronization distribution of the system at the same time.In summary,the method proposed in this paper can effectively perform correlation analysis and simultaneous measurement of semantic and non-semantic health care data,and provide a preliminary and effective method for data-driven health care decision-making.