DRAFT DRAFT DRAFT DRAFT DRAFT DRAFT DRAFT DRAFT DRAFT
Object-orientation in software development is not a new concept. However, the practical adoption of object technologies, particularly distributed object technologies, has been a slow process. In healthcare, fueled by the advances in internet, web and Java-based technologies, there has been a growing awareness of the need and usefulness of distributed object technologies. In this paper, we present the role object technologies play in healthcare, with a particular focus on the various standards bodies, followed by sample application of these technologies in practice.
Though there are innumerable standards in healthcare, a distinct few have recently achieved a fair degree of prominence. Highlighted here are a few key standards-making efforts that are relevant to the use of evolving object and internet technologies for healthcare. In the following figure, we compare the focus of some of these efforts. The comparison is presented in terms of client-server environments. Later we look at these from an n-tier architectural viewpoint and discuss the role of object technologies in middleware development.
The Object Management Group (OMG) is a not-for-profit entity engaged in developing standards for system integration using object technology. OMG is the largest software consortium in the world with over 800 member companies. OMG's Common Object Request Broker Architecture (CORBA) standard has brought interoperability to distributed object computing [See: CORBA]. As its next set of targets, the OMG has formed a number of task forces in vertical market segments. The CORBAmed Task Force is OMG's effort in the healthcare domain. CORBAmed is actively creating interoperable interface specifications for distributed object services (e.g. Patient Identification Service, Lexicon Query Service, Clinical Observation Service) which will be made commercially available from various vendors.
OMG has an interesting and proven approach to standardizing technology that is intrinsically different from the conventional standardization process. It brings together component technology vendors and users in an open forum, challenges them to identify areas that require standardization that are both practical and realizable, publicizes such areas widely, requests and collects proposals from competing vendors, drives for a consensus among them, and adopts the result. One interesting aspect of this process is that an officer of any company responding to an OMG Request for Proposals must agree to commercialize the standard within a year. This process has resulted in consensus standards in areas which have defied any standardization in the past, with the added benefit of assured commercial product availability.
The CORBAmed task force has a very vigorous following and is developing service specifications in numerous areas in healthcare - leveraging existing standards such as HL7, DICOM and others.
The focus of CORBAmed is on interoperable services - leveraging the strength of CORBA. This is in effect a focus on backend technologies and middleware technologies.[See: CORBAmed]
The Clinical Context Object Working group (CCOW) is developing interoperable solutions for application integration on the desktop. The applications need not be limited to the desktop; they may be client-server. However, integration is via information which is presented and/or entered at the client side, and all such applications must run in windows on the same desktop environment. For example, CCOW may be used to capture a new patient entry into a registration program and use this to trigger a managed care program to retrieve patient eligibility and payment information (as if the patient had been manually selected on the latter). CCOW is oriented toward Microsoft's Active-X/COM/DCOM environment [See: DCOM], however, intergration via CORBA is an option. The specifications that this group is currently in the process of developing address how multiple client-end applications from multiple vendors communicate and switch contexts interoperably in the desktop. [See: CCOW]
HL-7 stands for Health Level Seven. HL7 is used to denote both a data messaging standard for the exchange for healthcare information and is an ANSI accredited standards body which has focussed extensively on data formats for exchange within organizations. Founded in 1986, HL7 now has more than 1500 members. This group is the main source of content information and healthcare models in the U. S. [See: HL7]. HL7 members have long been interested in adopting and promoting object technologies. There is a serious effort underway with the HL7 version 3.0 Reference Information Model (RIM) to use object-oriented techniques.
One subgroup in HL7 is looking at Object technology in general and Object Brokering technologies in particular, and is called the SigOBT group. The efforts of this group has been the source of content for the Andover Working Group and ActiveX for Healthcare consortiums discussed below. [See: HL7 SIGOBT]
This is a relatively newer effort started about a year ago focusing on content tagging using XML. The current focus of this group is translation of the current HL7 ascii encoding by reformatting with XML tagging. This group also intends to adopt the new versions of HL7, which takes a more object-oriented view.
The Andover Working Group is a vendor consortium devoted to providing solutions based upon standards such as CORBA and Active-X. The AWG has developed Enterprise Communicator software which "wraps" HL7 messages with object oriented transport mechanisms, obviating the need for socket-level programming. The AWG is currently looking at imaging and the DICOM standard (below) to see what developments can be useful to its members.
This group's goal is to promote the adoption of Microsoft's platform technologies in healthcare. Initial focus of the group has been similar to the Andover Working Group, but focuses on the development of ActiveX-based implementation of HL7 messaging objects.
The DICOM standard (formally known as ACR/NEMA standard) focuses on the exchange of medical images (e.g. digitized x-ray, CT, MR, ultrasound, nuclear medicine, angiography, etc.) and related data. DICOM Structured Reporting Supplement 23 presents a model for relating physical entities (e.g. tissue samples) with images and with observations/measurements made on those images. [See: DICOM]
The Association for Standards in Testing and Materials (ASTM) group E31 develops "high level" standards for healthcare. These standards focus on a "higher level" of applicability; e.g., what data should be in an Emergency Room clinical record.
The X12 Committee develops international standards for the exchange of billing information. They, too, see the future in terms of object technology and are adding XML tags to existing data elements.
The IEEE Measurement Image Bus (MIB) standard defines a protocol for the exchange of measurement data from clinical instruments (patient monitors, infusion pumps, etc.).
The efforts listed above are but a few examples of the "zoo" of standards organizations that exist in health care. The sheer quantity of the standards they generate has led many health care IT professionals to view standards in general with some degree of skepticism. In reality, however, there is not a great deal of overlap between or among these standards, and the people who contribute to their development generally try hard to make use of the best features of existing standards, rather than compete with them. The figure at the top of this paper illustrates how several of these standards play together in a distributed healthcare information environment.
HL-7 is a messaging standard. It is used to exchange healthcare related information between two entities via a coded textual message. The messages may be sent ad-hoc, e.g. based upon certain trigger events in the sending system, or the messages may be sent in response to a query (however, the latter mode is not well supported at present). In its current utilization, HL-7 works best for exchanging information between host systems, where each host needs a duplicate copy of the information (e.g. patient demographic data). In the past, HL-7 has been criticized for a lack of an information model to standardize the semantics of the information being exchanged. Thus, many "flavors" or interpretations of the standard exist in the real world today. However, HL-7 is moving to rectify this problem in version 3.0.
CORBAmed is based upon the CORBA distributed object model. CORBA provides a robust and highly scalable infrastructure for a distributed object model of computing. The CORBAmed domain task force is chartered with developing standards for the "business objects" for health care. By viewing a set of hosts as a network of distributed, communicating "objects", clients are able to obtain processed information over an enterprise-scale network. In the longer term, hosts may utilize each other's distributed object services to provide information on demand, thus providing an alternative to the exchange of HL-7 messages. However, in the short term, the efforts of CORBAmed will best be utilized in client-server communications. By providing an object-oriented view of clinical information, CORBAmed allows hosts to "protect" their data by exposing only specific "methods" to clients. Clients can rely on CORBA hosts to provide information on demand and need not keep duplicates of information in local storage, obviating problems of freshness and authenticity which are inherent in message exchanges. CORBA standards are platform and language neutral, which is a key requirement for any large scale client-server integration effort.
As a distributed object technology, CORBAmed must adopt an object model for its domain. The HL-7 version 3 modeling efforts are being closely looked at for guidance in this regard. In addition, real world HL-7 experience will be used in the definition of clinical observation information to be exchanged via the forthcoming CORBAmed Clinical Observation Access Service (COAS). In fact, the COAS RFP mandates that existing standards (including HL-7, DICOM, MIB and others) be looked at for definitions of the information to be exchanged. Furthermore, CORBAmed has an active working group looking at interoperability with HL-7.
The Andover Working Group (AWG) is an ad-hoc collection of vendors who wish to use HL-7, DICOM and other message exchange standards "wrapped" in distributed object technology (CORBA and DCOM). AHC has goals similar to Andover but focuses exclusively on Microsoft's DCOM and ActiveX technologies. By "wrapping" an HL-7 message in CORBA, a vendor inherits the benefits of a robust "bus" for finding, securing and exchanging information in a heterogeneous computing environment. "Wrapping" methodology does not provide the full benefits of object orientation (i.e. it does not support inheritance or polymorphism) but it does utilize robust off-the-shelf "middleware" for communication and infrastructure services, thus lowering development costs and ensuring interfacability between heterogeneous systems.
CCOW is an outgrowth of the Microsoft Healthcare Users Group (MHUG). As such, it uses Microsoft's object technology (COM, DCOM) rather than CORBA (although a CORBA solution is allowed). Because of this, many people assume that CCOW and CORBAmed are direct competitors. In reality, they are not. Although DCOM may some day compete with CORBA in the enterprise-wide distributed object world, the current focus of CCOW is where Microsoft is strongest today - i.e. on the desktop. CCOW members are vendors of Windows-based desktop applications who wish to provide their customers with a measure of automation via Active-X controls and COM. This is a very powerful mechanism for helping users to automate their desktop applications. Although CORBA could be used for the same purpose, Active-X is most closely tied to the Windows environment and provides a very rich set of inter-application controls and interfaces. Therefore, CCOW is aimed at desktop (client) automation, and CORBAmed is aimed at client-server interfacing.
The previous discussions basically point to two major trends in healthcare in the utilization of objects. One methodology focuses on the use of messaging from HL7 to AWG to AHC, so some call it Message Oriented Middleware (MOM) -. The other trend focuses on middleware services, exemplified by the CORBAmed approach. Each of the approaches has a role to play, depending on the context, desires and goals of vendors and customers. Messaging using HL7 in healthcare is typically characterized by monolithic applications sharing content that gets duplicated in each application. The figure below exemplifies this philosophy.
Messaging has an important role to play when used properly. For instance, it is the technology of choice to notify physicians or other stake-holders on important or changing information. However, messaging used simply to preserve the status quo or to protect a particular vendor's existing technologies will simply delay the reaping of benefits afforded by internet and distributed object technologies.
HL7, in its version 3.0, has focused on developing a comprehensive reference information model that serves as a blueprint for designing middleware in healthcare. These models have been a fertile source for both message-oriented and n-tier-oriented middleware service developments.
From the perspective of users of technologies such as hospitals and clinics, adoption of component technology is driven by the promise of simplification, primarily in the management of the applications and the reduction in duplication of functionality. HL-7 and CCOW are vendor-centric approaches that address primarily the exchange of information, (now done with objects in HL7 v3.0) between existing systems and are not aimed at reducing duplication in functionality nor simplification of the management of these systems. Object technologies and frameworks such as the Object Management Architecture (OMA) and Microsoft's DNA, provide the promise of true reuse, enterprise-wide deployability and a simpler maintenance and evolution path for users. In addition, object technologies facilitate support for processes in organizations. When processes must be changed to fit technology constraints, the acceptance and benefits of any new technology can be reduced dramatically. By breaking process into discrete components, it is easier to support the changes in individual processes that are needed to allow technology to succeed. Eventually there are likely to be a number of standardized component technologies in healthcare. However, one does not have to wait for these standards to be fully developed to realize the benefits component technologies can make possible today. The primary advantage of using standards-based components in healthcare information system architectures is that any standards-based component can be seamlessly and inexpensively replaced at any time with a later generation, improved, standard-based component while the rest of the processes and technologies involved can remain intact and unchanged5.
Both CareFlow|Net, Inc. and Philips Medical Systems have developed and demonstrated the use of component and object technologies for supporting various healthcare processes at Baptist Health Systems of South Florida (BHSSF). CareFlow|Net, Inc. has deployed an Enterprise Transcription Solution at BHSSF, based completely on object technologies from CORBA to Java to ActiveX.
Conclusions
While it is true that the many standards activities within the health care arena are not perfectly aligned, they are generally not competitive, either. Health care IT professionals are not so much faced with a choice of standards, but rather a set of standards to use in various, different circumstances. HL-7 will continue to dominate in the host - host world, although CORBA could play a role here as well, particularly in situations where the retention of data ownership rights is an issue. CORBAmed should provide the standards for client - host computing, as CORBA middleware is robust and proven in enterprise-scale applications. CCOW should continue to focus on desktop automation. DICOM should provide the model for observational reporting as well as the representation for medical image data, particularly for primary diagnosis. Likewise, MIB should continue to focus on instrumentation interfaces, ASTM on health care applications standards (computer-based patient records, basic emergency data set, etc.), ASC X12 on billing standards, and various government and private organizations on coding and vocabulary standards. Users, in the meantime, should be represented on all of these standards-making bodies and should insist that they continue to cooperate toward greater re-use and interoperability.
V. "Juggy" Jagannathan, Ph.D. is the Associate Director of Research at the Concurrent Engineering Research Center (CERC) and Associate Professor of Computer Science. He is the Co-PI for the two National Library of Medicine sponsored projects on healthcare at CERC, funded at close to $8 Million for six years (1993-1999). He is also the Sr. VP of R&D and CTO of CareFlow|Net, Inc., a spin-off company to commercialize the ARTEMIS technology developed at West Virginia University. Dr. Jagannathan has over 60 publications in workshops, conferences, books and journals. Dr. Jagannathan obtained a Ph.D. in Electrical and Biomedical Engineering from Vanderbilt University in 1981.
Kent Wreder, M.S., has directed operations and project management in the healthcare informatics field for over 5 years. As Corporate Director of Object Technology for Baptist Health Systems of South Florida (BHSSF), he is responsible for systems architecture and in-house software development for the BHSSF computer-based patient record (CPR). With an organization that includes five hospitals in the Miami area, BHSSF is the largest not-for-profit integrated delivery system in southern Florida. Prior to his employment with BHSSF, Mr. Wreder served as research engineer in the Corporate Research Department at Cordis Corporation in (Miami, FL)), where he was employed for 3 years. His primary area of responsibility included feasibility study and design of medical devices such as balloon angioplasty, laser angioplasty and intravascular ultrasound. He also participated as research engineer in the project team that researched radial based artificial neural networks for stereotactic radiosurgery under a research grant through the Neuroscience Center at Baptist Hospital of Miami and Florida International University. Mr. Wreder received his B.S. in Mechanical Engineering and his M.S. in Computer Engineering from Florida International University, Miami. Currently, he is completing coursework towards a Ph.D. in Computer Science at the same university.
Robert. Glicksman, MSEE, is currently the Director of the MedGRiD program with Philips Medical Systems. In this capacity, he oversees all aspects of a program to integrate clinical information, including business development, project management, engineering and program operations. Prior to this, he was the program's Chief Scientist and led the development of a three tier system architecture based upon distributed object technology (CORBA) and Java. He has 30 years of engineering experience in signal and image processing, image management, image distribution, image storage and image compression. He has worked on medical imaging applications for the past 18 years, including teleradiology, Picture Archiving and Communication Systems (PACS), and interfaces to healthcare information systems. He holds MSEE and BSEE degrees, has one patent, and has co-authored numerous papers on system architectures for PACS and enterprise information systems for healthcare.
Yasser alSafadi, Ph.D, received his degree from the Electrical and Computer Engineering Department at the University of Arizona. Currently, he is a Senior Member of the Research Staff at Philips Research. He is leading the project on Healthcare Interoperability, which builds testbeds for experimenting with OMG's CORBA, and with Microsoft's DCOM object brokering technologies and eXtensible Markup Language (XML). His research interests are in: Distributed component architectures for interoperable systems, Multimedia servers and service delivery. He co-chairs the Special Interest Groups: Health Level 7 Image Management, Andover Working Group (AWG) for Open Healthcare, Interoperability - DICOM, CORBAmed BioMedical Imaging Working Group.
Associate Professor, Associate Director, CERC, West Virginia University and Sr. VP R&D and CTO of CareFlow|Net, Inc.
Corporate Director, Object Technology, Information Technology Department, Baptist Health Systems of South Florida
Director, MedGRid Program, Philips Medical Systems
Senior Member of Research Staff at Philips Research
If object oriented component technologies had always been used in information technology, there would be only one core date/time class and the Y2K problem would not exist!
