Demographic management in a federated healthcare environment

https://doi.org/10.1016/j.ijmedinf.2006.04.006Get rights and content

Abstract

Objective

The purpose of this paper is to provide a further step toward the decentralization of identification and demographic information about persons by solving issues related to the integration of demographic agents in a federated healthcare environment. The aim is to identify a particular person in every system of a federation and to obtain a unified view of his/her demographic information stored in different locations.

Method

This work is based on semantic models and techniques, and pursues the reconciliation of several current standardization works including ITU-T's Open Distributed Processing, CEN's prEN 12967, OpenEHR's dual and reference models, CEN's General Purpose Information Components and CORBAmed's PID service.

Results

We propose a new paradigm for the management of person identification and demographic data, based on the development of an open architecture of specialized distributed components together with the incorporation of techniques for the efficient management of domain ontologies, in order to have a federated demographic service. This new service enhances previous correlation solutions sharing ideas with different standards and domains like semantic techniques and database systems. The federation philosophy enforces us to devise solutions to the semantic, functional and instance incompatibilities in our approach.

Conclusion

Although this work is based on several models and standards, we have improved them by combining their contributions and developing a federated architecture that does not require the centralization of demographic information. The solution is thus a good approach to face integration problems and the applied methodology can be easily extended to other tasks involved in the healthcare organization.

Introduction

Throughout everybody's lifetime, any person may have episodes of care assisted by many health-care professionals. This is mainly due to the increasingly specialization of medicine practice and to the growth of population mobility, among other factors. On the other hand, new information and communication technologies (ICT) are being applied in healthcare tasks making these easier and more efficient. The extensive use of ICT is having a clear impact in the transition to the digital hospital [1] being the management of multimedia clinical information systems an essential issue.

In this new scenario, patient's clinical information is distributed among several information systems, geographically dispersed. Usually, these are autonomous and heterogeneous systems developed under different objectives, and consequently with diverse data models, platforms, standards and semantics. In the best conditions, they could follow a standard, but different ones could be selected for each system. The compilation, visualization and management of the distributed clinical information related to a patient can ease the healthcare professional tasks and improve the quality of service. The query of this information to give a single, coherent and contextualized record of the EHR needs that these legacy systems work in a collaborative environment. When the autonomy of the systems has to be maintained collaboration is needed in a federated approach [2], [3], [4] and the resolution of incompatibilities between the systems of the federation must be accomplished. These conflicts can be classified into three levels [5], [6], [7], [8], [9], [10], [11]:

  • Semantic: the information models or database schemas are different.

  • Functional: the interfaces to manage data are different.

  • Instance: the information about the same real entity could differ from one system to another [12].

The large number of applications, mutually isolated and incompatible, that are already available in the market, installed, and operational in healthcare organizations, effectively supporting specific needs of users cannot be underestimated. Under the present circumstances, a critical issue is to make possible the integration and interoperability of already existing applications thereby securing investments already made and allowing continuity of the service whilst facilitating a gradual migration of existing proprietary, monolithic systems towards the new concepts of openness and modularity. The cost-effectiveness of the solutions, especially when projected on the scale of the whole organization, represents another crucial aspect to be evaluated carefully [13]. Several groups and standard organizations are making efforts to solve these integration and interoperability problems [14], [15], [16], [17].

In order to collect the patient's healthcare information within an autonomous system from the federation, the patient must be uniquely identified first. Usually a system will allow the user to submit a search for a patient's record using some combination of identity parameters for that person [18] and when the identification is finished this is translated into a unique identifier (PersonID) used in the whole system to reference this patient.

The assignment and maintenance of PersonIDs is frequently local to the system. This management style suits the purposes of recording and retrieval of patient's records inside the local domain. However, the efficient collection or correlation of health records in a federated context requires more sophisticated identification and PersonID management processes.

For the management of EHR in a federated scenario the person has to be correctly identified in every system. An obstacle for this identification is the lack of a unique person identifier. Some administrative identifiers, as Social Security Number, National Identity Document, Passport Number, or Car Driver License could be considered, but systems usually manage a local PersonID independent from these administrative identifiers because there is no guarantee about the existence of them for every registered person [19]. Especially in international scenarios, where many different administrative domains are involved, this difficulty increases. To keep the autonomy of the systems, this consideration has to be maintained when an open integration solution is developed.

After the identification of the person, the demographic data about the patient has to be requested over the federation in order to have a unique view of his/her demographic knowledge. Only after these first tasks, which are carried out by demographic components, the whole EHR would be visualized [20]. The importance of these components, which objective is to manage the identification of a person and the query of his/her demographic data, is thus justified.

In addition, it is important to emphasize that, if the developed methods are not centered in the patient but in the identification of every person involved in the healthcare organization, and if a well designed component-oriented approximation is adopted, the results could be reused in different activities inside the organization, for example in security or healthcare staff management, and not only in those directly related to the EHR management.

When demographic components are designed independently, they could have different database schemas and could manage different concepts and data types. Interoperability is only possible if previously they are agreeing in a common semantic of the managed concepts. This agreement could be established in the specification of an ontology, which describes the domain concepts and the relation among them.

For example, one system can store the age of a person as “age”, another can manage “birth date” and a third one can use the Spanish term “edad” to represent this concept. The common ontology should include the concepts “age” and “birth date”, and the relation between them. This way, the dialogue between two different components that manage these two concepts will be possible, understanding the information recovered from each one and facilitating the integration of this information. A component that understood the “edad” concept has to make the mapping to the “age” concept when it is interacting in federated workflows. Ontology management methods could simplify the management of relations between concepts and the mapping from local to federated concepts.

In our framework the common ontology to describe the demographic knowledge is based on CEN [21], [22], OpenEHR [23] and CORBAmed Person Identification Service (PIDS) [18] information models, which are described in Section 2. We use OWL as ontology language because it is the standard proposed by the Web 3 Consortium (W3C) [24], [25] and thus the reusability of our solution will be improved. Several research groups in medical informatics are using this language in their works [26], [27], [28], [29].

When demographic components are designed independently, they usually have different functionalities and interfaces. However, in a federated environment, the unique identification and view of the demographic knowledge about a patient are essential. This requirement implies the establishment of federated workflows to solve integration problems, maintaining the distribution of the information, and the autonomy of each system in the federation.

The management of identification and demographic knowledge requires the standardization of interfaces. Each autonomous system participating in the federation must export its functionality according to the normalized interfaces. The concepts exchanged through the interfaces are described in the common ontology developed for semantic integration.

To accomplish the objective of integration, we have used the interfaces described in the PID service [18], detailed in Section 2, as a starting line, extending them to configure a more decentralized approach and to facilitate the semantic conflict resolution, as it is explained in Section 3.

When demographic information about the same person stored in different systems has to be merged, another two problems need to be accounted for. Concerning the person identification, the corresponding PersonID handled by each system must be found, with the guarantee that it refers to the correct person. On the other hand, it is possible that the demographic attribute values of the same person in different systems do not match, and value conflict resolution arises. For example if the address takes different values in two systems, a mechanism to choose the correct value has to be implemented.

Considering that the systems in the federation continue supporting their pre-existing functionality in addition to the new federated one, instance incompatibility can be a recurrent issue once the previous problems (semantic and functional) have been solved.

In this paper, both unique identification and demographic data queries are addressed. The main objective is to get an open and optimal specification for demographic servers in a federated environment. By an open specification we mean an extensible method, independent from specific technological implementations. This objective is concreted into an architecture of components for which the integration of new systems inside the federation is an easy task and does not involve changes in the management of identifiers and demographic information. Two other important issues to provide an optimal specification are the cost related to the exchange of patient data, which should be minimal, and the view of the patient demographic information in the overall system, which must always be updated.

Our approach accomplishes the decentralization of identification and demographic information about persons. It represents a further step toward the solution of federated identification [2], [18], compared with previous widely known efforts in this line, like the Person Master Index (PMI) [30], [31] or Unique Person Index (UPI) [32], which are based on the centralization of person data.

We propose a new paradigm for the management of PersonIDs and demographic data in order to provide a PMI service. This new service enhances correlation solutions in the PID service from CORBAmed [18] sharing ideas with other standards and domains like semantic techniques and database systems. The federation philosophy forces us to propose solutions to the semantic, functional and instance incompatibilities in our approach.

Section snippets

Methods

This work is based on a component architecture modeling [13], [33], and on contributions taken from three main sources: OpenEHR dual and reference models [23], [34], [35], CEN's GPICs (General Purpose Information Components) [21], [36] and CORBAmed's PID service [18].

As a starting point, we use the PID service to describe the demographic service interface and solve functional-level incompatibilities. To account for the special needs of federated environments, PID service interfaces have to be

Results

Using the methods presented in Section 2, we have designed a component's architecture for the management of person's identifiers and demographic information. As any architectural solution within the European context should be, it is compliant with prEN 12967-1, 2, 3 and, consequently, with ODP methodology.

The computational viewpoint is based on PID service interfaces, although these have been extended with some useful functions that improve the correlation tasks in order to offer a more

Discussion

Person identification and demographic data management are essential tasks in the activities of a healthcare organization, in special in those involved in the management of the EHR. Our work is focused on an approach that faces the identification of persons and the recovery of their demographic knowledge over heterogeneous clinical information systems. The integration problems for semantic and functional incompatibilities are solved and the value conflict resolution techniques can be deployed in

Acknowledgments

We thank the reviewers of the journal for the comments to clarify issues in the questionnaire survey and make the paper more structural and readable. The authors also wish to acknowledge the contributions of the OpenEHR members as thoughts and problems about this work have been exposed in their distribution mailing lists. These collaboration efforts are very important for standardization processes. This work was partly supported by the Spanish Ministry of Science and Technology

References (41)

  • E.-P. Lim, J. Srivastava, Entity identification in database integration: an evidential reasoning approach....
  • E.P. Lim et al.

    Entity identification in database integration

  • E.-P. Lim et al.

    An evidential reasoning approach to attribute value conflict resolution in database integration

    IEEE Trans. Knowl. Data Eng.

    (1996)
  • E.-P. Lim et al.

    Resolving attribute incompatibility in database integration: an evidential reasoning approach

  • J. Parsons et al.

    Attribute-based semantic reconciliation of multiple data sources

    J. Data Seman.

    (2003)
  • J. Parsons et al.

    Emancipating instances from the tyranny of classes in information modeling

    ACM Trans. Database Syst.

    (2000)
  • CEN/TC 251.Secretariat: SIS, Health Informatics–Service Architecture (HISA). prEN 12967-1. Part 1. Enterprise...
  • OMG, Healthcare Domain Task Force standards web site. URL: http://www.healthcare.omg.org/Roadmap/corbamed_roadmap.htm....
  • CEN/TC 251 web site. URL: http://www.centc251.org. Note: last access December...
  • HL7 web site. URL: http://www.hl7.org. Note: last access December...
  • Cited by (20)

    • Integrating reasoning and clinical archetypes using OWL ontologies and SWRL rules

      2011, Journal of Biomedical Informatics
      Citation Excerpt :

      Note that both trees are mixed in a way that remains compatible with the original RM structure. The point of departure is the mapping of the openEHR RM to OWL developed by Román et al. [27]. Archetype definitions start with an ENTRY subtype like EVALUATION, INSTRUCTION, ACTION or OBSERVATION.

    • Investigating telemedicine approaches: A 10-country comparison

      2018, International Journal of Engineering and Technology(UAE)
    • The design of flexible telemedicine framework for healthcare big data

      2018, International Journal of Engineering and Technology(UAE)
    View all citing articles on Scopus
    View full text