HealtheVet Reporting Engine, 10-R9646Printer Friendly Version
Inclusive Dates: 07/18/06 11/20/06
Background - SwRI has worked with the Veterans Health Administration for several years on numerous projects. Recently, the process of re-hosting legacy systems to a service-oriented architecture (SOA) has been under way. In an SOA, services have defined interfaces through which other applications (internal or external to the system) may communicate. In these legacy systems, there are a multitude of existing services. Specifics of the associated interfaces are essentially indeterminate. The current reporting tools and methods in use are not consistent with an SOA. Additionally, report mechanisms on the re-hosted systems had been developed through the manual process of obtaining and analyzing documentation to determine the interface from each service. Hence, a need for an automated means of discovery of service metadata to support future reporting capabilities was apparent.
Approach - This project was undertaken to prepare for a future VHA proposal for the required automated discovery of data service metadata. The first area of focus for this project was to obtain an understanding of the legacy architecture. This was accomplished through research of the available documentation for the existing Health Information Model (HIM) and analysis of the existing legacy systems and recently implemented re-hosted applications.
The second area of focus was the development of design options for dynamic discovery of data service metadata. Potential designs using universal description, discovery, and integration and web service definition language metadata discovery were documented and reviewed. As the project progressed, additional techniques such as the Massachusetts General Hospital Utility Multi-Programming System remote procedure call and direct HIM access were investigated. The resultant potential designs for sample re-hosted systems were further refined into detailed design concepts.
Accomplishments - Approximately 30 domain packages within the HIM were discovered, analyzed, and summarized, including a timeline of domains that have yet to be implemented. Duplicate metadata structures were found in the HIM, possibly resulting from a requirement to be aligned with the Health Level Seven (HL7) Reference Information Model. Example mappings between each were produced. Referential and data integrity constraints that exceed the capability of HL7, UML, Java, and XML were also discovered within the HIM.
Two design options were developed and documented, along with their associated risks. The first design option would interface directly with the existing legacy applications. This design has the potential to provide all the metadata that is needed by the reporting engine, thereby minimizing required development for a presentation layer to interface with each re-hosted application. The second option details a reporting engine design that interfaces with both application services and data services. In this approach the reporting engine would contain a presentation layer, business layer, and data layer. The business layer would implement all the business rules required for dynamically discovering the metadata. The data structures based on the HIM domain packages would be passed between the business layer and the data layer. While more complex, this second and recommended design option is robust and provides for dynamic data discovery. The immediate opportunity for a VHA proposal was eliminated due to VHA leadership and budget revisions. Results of this IR&D were reflected in a white paper delivered to the VHA for potential future funding consideration.