| With the continuous development of medical informatization,more and more clinical data are stored in the form of electronic health records.The use of clinical data in electronic health records to conduct clinical research can greatly improve the efficiency of clinical research,but how to quickly select patients who meet the eligibility criteria for clinical research from the massive electronic health records database is the key.Traditionally,patient screening in clinical research is often done based on general data query tools.The general data query tool is easy to operate by screening patients through the “concept” + “constraint” form.However,due to the complexity of the eligibility criteria,it has issues that the concept is not completely matched with the electronic health record data items,and the constraint conditions cannot by fully expressed.Moreover,the actual conversion to SQL to query the electronic health record database involves multiple layers of sub-queries and join queries,resulting in inefficient execution performance.In response to the above issues,this thesis studies the computer-processable expression technology and execution technology of eligibility criteria,and on this basis,realizes a highperformance clinical research patient screening tool.The main research contents and innovations are as follows:1)By analyzing 4,691 eligibility criteria in 389 clinical research related to stroke,it is concluded that there are three matching forms between concepts and data items: direct mapping,calculation mapping,and rule mapping,and there are six types of constraints: existential,comparative,logical,sequential,causal,and collective.We take the i2b2 query tool as an example to analyze the advantages and limitations of the general data query tool in the expression and execution of the patients screening conditions of clinical research.2)In view of the difficulties in mapping between concepts and data items and the incomplete expression of constraints in general data query tools,the computer-processable expression of eligibility criteria is studied.We propose a screening condition expression method based on the openEHR model and openEHR EL,which realizes the multi-level mapping of screening concepts and electronic medical record data items and realizes the constraint expression of eligibility criteria through the openEHR EL expression language.Through the expression evaluation of the actual eligibility criteria,the results show that the screening condition expression method is extensible and can express more eligibility criteria than traditional expression methods.3)Aiming at the problem of inefficient performance of general data query tools for clinical research patient screening,the implementation technology of clinical research screening conditions is investigated.We propose a screening execution engine based on the universal search engine elasticsearch,which specifically includes: the elasticsearch index mapping automatic generation method based on the open EHR model,which realizes the middle-tier storage of electronic medical records;the open EHR EL interpreter and translator,which translate the screening conditions as elasticsearch query statements for execution,etc.The performance of the filter is evaluated through real filter conditions,and the experimental results show that the eligibility criteria execution engine has better performance.4)Designed and implemented a clinical research patient screening tool.We evaluated its application in actual clinical research.The evaluation results show that using the screening tool,researchers can customize concepts and constraints according to the needs of the research topic with flexibility and scalability,which can meet the needs of patient screening and reduce the burden on researchers. |
