Design And Implementation Of Clinical Data Repository Based On OpenEHR

Clinical Data Repository (CDR) is a clinical core system that provides unified management and open access of clinical data that have been dispersed in different clinical information systems, e.g. electronic medical record system, computerized physician order entry system, laboratory information system, picture archiving and communication system, etc. Most of the CDR are designed and developed using a "single level" approach, in which information model and domain knowledge are modelled directly in software and database models. Such CDR systems are usually large scale and difficult to maintain. With the medical knowledge and requirements changing, the model of the CDR has to be modified and re-designed. Frequent change will eventually cost a lot of manpower and resources.To address above problems, this paper uses a "two-level" modeling approach, which the openEHR specification proposes to separate the information and knowledge, to design and develop CDR. Developers implement the system based on the reference model, while domain experts define domain models to describe clinical knowledge and requirements. The CDR system can transform the domain models to information model, storage structure and access interface automatically. The functions of the CDR system are driven by the domain models. Thus, the change of the clinical requirements can be solved by changing domain models instead of re-designing and developing the entire CDR system.The main work of this paper include:1) Analyze the openEHR "two-level" modeling approach and health computing platform (including reference model, archetype, template, archetype query language, etc.). Research the application of openEHR specification in relevant aspects and propose our approach on designing and developing CDR.2) Research the storage solution of the openEHR specification. Propose the Template Relational Mapping (TRM) method based on the Archetype Relational Mapping (ARM) method to transform domain models to database schemas automatically, which makes the underlying relation and data structure transparent to domain experts so that they can focus on defining and completing domain models.3) Research the Archetype Query Language (AQL) and its drawback. Propose the Template Query Language (TQL) based on AQL to provide model-based data query, which empowers the clinicians to query clinical data based on the domain models without knowing the underlying data structure. And TQL also provides SQL-like flexible query condition, e.g. and, or, matches and brackets.4) Design and develop the CDR system based on the data storage approach in 2) and the data access approach in 3). The CDR system also provides the management of the model’s entire lifecycle so that the creation, modification, submission, approvement and deployment of the model can all be done on the system.The application of the CDR system in two designed scenarios proves that the proposed approach to design and develop CDR can meet the growing and changing clinical requirements well. In addition, the CDR system has already been applied in one of grade 3 A hospitals in China. The models that the domain experts defined have been transformed into database schemas successfully and applications, e.g. medical integration view, can access the data of the CDR through the TQL interface.