Study On The Evaluation Technology Of Electronic Medical Record Data Quality Based On Rule Expression

The state issued a document pointed out that it is necessary to give full play to the role of electronic medical records informatization,use electronic medical records data to promote the improvement of medical management,improve medical service experience,and promote the development of smart hospitals.However,the data quality of electronic medical records will directly affect the effect of its secondary use.Therefore,it is essential to evaluate the quality of the data before using it.For structured electronic medical record data,the rule-based data quality assessment method is the most widely used.This method requires the assessor to transform the data quality assessment requirements into computer-executable rules.The current method of using query language to define evaluation rules requires evaluators to master and strictly follow the corresponding grammatical requirements,which is cumbersome to write and requires high labor and time costs.The method based on rule expression proposed by the existing research simplifies the rule definition process and reduces the technical requirements for evaluators through the structure definition of variables,functions,parameters and other structures.However,the following problems still exist: 1)the existing rule expressions are related to their respective research scenarios and lack a more general rule expression for the quality evaluation of electronic medical record data;2)there are many data items to be evaluated,which leads to a large workload of defining rules one by one.To address these issues,this study designed a set of more general electronic medical record data quality evaluation rule expressions to make the rule definition easier and faster.The method to automatically generate evaluation rules using the existing knowledge of the information model was proposed to reduce the workload of manually defining rules.On this basis,a data quality evaluation system was designed and implemented to provide an application framework for efficiently defining evaluation rules.Details of the thesis are as follows.(1)To simplify the rule definition process,twelve kinds of data quality evaluation rule expressions and two logical operators were designed.The regional health information platform is one of the important and typical application scenarios of electronic medical record data.The data quality assessment requirements of this scenario are clearly defined and representative.We analyzed data quality assessment requirements and the deficiencies of existing research according to its data collection specifications,extended the rule expression method based on the guidelines definition language(GDL),and used the Spark technology framework to achieve the rule execution engine.The proposed rule expression method allows users to define rules at a higher level.By comparing with actual requirements,the validity of the expression method was proved.(2)To reduce the workload of manually defining rules,a method of automatically generating evaluation rules using the existing data constraint knowledge in the information model was proposed.Taking the openEHR international standard information model as an example,the feasibility of the method was analyzed.Then the information model knowledge was analyzed in detail,and the knowledge related to data constraint requirements was extracted from it,and the mapping relationship between knowledge and rules was designed for each data type,and finally the automatic mapping of knowledge and rules was realized.(3)Designed and implemented a clinical data quality evaluation system.The system was designed based on the above-mentioned research,using the rule expression to build a rule editor to help users customize the rules,and a large number of rules could be automatically generated by simply configuring the clinical information model file through the rule automatic generation technology,and cooperated with the task management,execution and result management modules.The entire system has provided users with an application framework for efficiently defining evaluation rules.(4)A case study was carried out in the context of the construction of the Shanxi Province National Health and Medical Big Data Platform Project.We constructed the information model based on its data collection specifications,and automatically generated about 90% of the data quality evaluation rules through the method proposed in this study,and the rest were defined by rule expressions,and used the rules to evaluate the data quality of a hospital in Shanxi Province to help it find data quality problems early to improve the pass rate of data reporting.The results show that the method proposed in this study can help evaluators to quickly define data quality evaluation rules,and significantly reduce the workload of manually defining rules.