OBJECTIVE: For the purpose of establishing electronic health record (EHR), business-oriented health data distributed in different systems should be integrated to focus on individuals. This study is aimed at collecting health data items that are now nationally available in various health information systems, and harmonizing them by modeling and defining the data elements. METHODS: This study followed a bottom-up strategy in data standard development. Health data items were identified and collected by referring to national health service regulations, consulting domain experts and performing field investigations. Data items were classified and modeled based on recognized domain knowledge, information standards and specifications developed by standard development organizations (SDOs) of other countries. Data elements were extracted from data items and defined according to ISO/IEC 11179 -Metadata registries (MDR) and confirmed needs. RESULTS: 1588 data items were collected from 33 recording forms that have been used nationally in health services, and were classified with a conceptual data model that was composed of 7 super classes (healthcare clients, healthcare providers, birth registry, health event/act, healthcare process, death, and others) and 15 classes (person's identification, person's socio-demographic characteristics, address, communication, provider-organization, provider-individual, birth, health event/act, observation, procedure, drug and material administration, recommendation, evaluation, expenditure, death, others). By normalizing the concepts and representations of data items, data elements were derived and defined as the attributes of classes in the data model. Data items were specified as instances of corresponding data elements. CONCLUSIONS: A large number of health data have been collected nationwide but person's life-long health record is incomplete and inconsistent now. To integrate such massive quantity of health data from various sources, a conceptual data model was established to organize data items, avoiding conflicts and duplications in between. For data consistency, data elements should be extracted from the data items and defined as attributes of classes in the data model by choosing essential metadata attributes. Treating data items as instances of well defined data elements might make data in different contexts manageable and agreeable. To be semantically unambiguous, further study should be performed to deal with the standardization of detailed medical information, and perfect the approach of data harmonization. Copyright 2010 Elsevier Ireland Ltd. All rights reserved.
OBJECTIVE: For the purpose of establishing electronic health record (EHR), business-oriented health data distributed in different systems should be integrated to focus on individuals. This study is aimed at collecting health data items that are now nationally available in various health information systems, and harmonizing them by modeling and defining the data elements. METHODS: This study followed a bottom-up strategy in data standard development. Health data items were identified and collected by referring to national health service regulations, consulting domain experts and performing field investigations. Data items were classified and modeled based on recognized domain knowledge, information standards and specifications developed by standard development organizations (SDOs) of other countries. Data elements were extracted from data items and defined according to ISO/IEC 11179 -Metadata registries (MDR) and confirmed needs. RESULTS: 1588 data items were collected from 33 recording forms that have been used nationally in health services, and were classified with a conceptual data model that was composed of 7 super classes (healthcare clients, healthcare providers, birth registry, health event/act, healthcare process, death, and others) and 15 classes (person's identification, person's socio-demographic characteristics, address, communication, provider-organization, provider-individual, birth, health event/act, observation, procedure, drug and material administration, recommendation, evaluation, expenditure, death, others). By normalizing the concepts and representations of data items, data elements were derived and defined as the attributes of classes in the data model. Data items were specified as instances of corresponding data elements. CONCLUSIONS: A large number of health data have been collected nationwide but person's life-long health record is incomplete and inconsistent now. To integrate such massive quantity of health data from various sources, a conceptual data model was established to organize data items, avoiding conflicts and duplications in between. For data consistency, data elements should be extracted from the data items and defined as attributes of classes in the data model by choosing essential metadata attributes. Treating data items as instances of well defined data elements might make data in different contexts manageable and agreeable. To be semantically unambiguous, further study should be performed to deal with the standardization of detailed medical information, and perfect the approach of data harmonization. Copyright 2010 Elsevier Ireland Ltd. All rights reserved.
Authors: Yuan Xu; Ning Li; Mingshan Lu; Elijah Dixon; Robert P Myers; Rachel J Jolley; Hude Quan Journal: PLoS One Date: 2017-10-26 Impact factor: 3.240