Design Of Patient Identity Matching Method And Implementation Of EMPI System

The enterprise master patient index refers to a patient unique identifier across multiple medical information systems.The master patient index system ensures the accuracy and integrity of patient information distributed across different systems by creating and maintaining index relationships for patient identification information in each medical information system.For the same patient,there are a large number of inconsistencies in the primary indexes between different records of the system and between different systems,so the core task of constructing the master patient index system is the patient identity matching.The current patient identity matching method has the following problems:1)the accuracy of the deterministic algorithm based on the exact matching of the key segments is not high enough;2)the lack of information items in the patient record will have a greater impact on the calculation of the matching result.In view of the above problems,this paper carried out the improvement and design of the patient identity matching method,and realized the master patient index system on this basis.The specific research contents and main work of this thesis include:(1)An overview of existing patient identity matching methods and application scenarios.Firstly,the importance of patient unique identification in patient identity matching and the main category of patient unique identification are introduced.The patient identity matching method without patient unique identification is discussed,namely deterministic algorithm and probability algorithm,and the principle and limitations of the commonly used patient identity matching method are analyzed.Finally,the main application scenarios of the patient identity matching method in the process of constructing the main indexing system arc analyzed and summarized.(2)A patient identity matching method based on multi-field similarity calculation is designed.In order to solve the problem that the existing deterministic algorithm calculates the similarity of the field is single and cannot objectively reflect the similarity of the field,the accuracy of the deterministic algorithm is improved by selecting the appropriate string similarity algorithm for the field characteristics.Compared with other deterministic algorithms in different scenarios,the evaluation results show that the method has higher accuracy and recall rate.(3)A patient identity matching method based on the Expectation-maximization algorithm is designed.In the case of missing information,the application of the maximum expected algorithm reduces the impact of missing information items and weight settings on the matching results.The method avoids the subjective influence caused by artificially setting weights,and improves the accuracy of patient identity matching in the case of missing patient record information.Compared with other patient identity matching methods in different scenarios,the evaluation results show that the method has higher accuracy and recall rate when the information loss is higher.(4)The master patient index system is developed based on the above patient identity matching method.The system realizes the management of the master patient index through the functions of registration,query,merging,linking,splitting,etc.of the patient record,which can meet the requirements of deduplication and matching two actual scenarios.The system has been applied in the regional health information platform project,and can efficiently established cross-referencing indexes through matching and linking of similar records.