Research On Health Care Data Integration Model Facing Semantic Heterogeneity

With the deepening of medical and health information construction,digital transformation has led to an explosive growth of medical data maintained by various medical information management systems.It is estimated that by 2025,the amount of medical data will exceed 10 ZB.At present,medical informatization has been fully carried out in most hospitals and achieved some results,but less than 3% of the hospitals have realized data communication.In other words,the data generated by each system are independent from each other,so it is difficult to communicate and share among the systems,and the phenomenon of "data island" becomes increasingly serious.Therefore,through the integration of multi-source and heterogeneous medical data,data sharing within and among medical institutions gradually becomes the key to realize the construction of medical informatization.Different from other industry information system data integration,there are many types of interdisciplinary subjects in the medical and health field,which involves complicated professional content,more difficult data integration,difficulty in system update synchronization,different expressions,complicated exchange methods,unstructured data is difficult to use and so on,makes the present stage the traditional ETL tool support to medical data integration of data warehouse technology is hard to meet the medical needs.Based on the data integration process in the construction of intelligent hospital,this paper conducts relevant research on the existing data integration problems to be solved.The main problems are:(1)In order to solve the problem of medical text information extraction,an algorithm based on bidirectional maximum matching of medical text segmentation technology was proposed,and an entity extraction combination more suitable for the medical field was designed.(2)In order to prevent semantic heterogeneity,the ontology-based synthetic similarity algorithm and MSA similarity detection algorithm are proposed to eliminate the semantic heterogeneity that is omitted due to one-sided evaluation and improve the accuracy of matching.(3)In order to improve the quality of data warehouse integration,the distributed ontology conceptual model oriented to the medical field was constructed in combination with the characteristics of medical heterogeneous data to be integrated and the business data requirements of medical institutions,so as to solve the problem of medical data heterogeneity,guide the ETL work and complete the construction of medical data warehouse.To sum up,this paper mainly focuses on the research and improvement of related technologies in the data integration stage of intelligent hospital construction.Combined with Chinese word segmentation related technologies,a named entity recognition scheme based on medical text was developed to extract medical text data information,improving the storage and integration efficiency of medical text information.Integrating medical data of various business systems,combining ontology theory and related mapping algorithms,made medical data integration more efficient and accurate,and facilitates the sharing and sharing of heterogeneous data between business systems.Optimize the data integration process of smart hospitals to provide data support for a variety of business needs such as in-hospital diagnosis and treatment,disease prediction,and decision analysis,while improving the efficiency of medical data resource management and sharing.