Decision support systems: fact and fiction.

Computers perform tasks daily that aid people in the completion of their duties; computers help drive cars, cook food, entertain, and support our decision making processes. Often a silent partner, the computer is involved in tasks that a health professional provides. Modern ECG machines, infusion pumps, blood analyzers, ventilators, and imaging machines contain computers that aid in the use of these devices. Though not traditionally considered decision support systems, these diagnostic systems support the decision making process in health care. Shortliffe EH et al define the kinds of decision support systems by the roles they play in the decision making process. The authors state that any computer system that deals with clinical data or medical knowledge is intended to provide decision support. This article will discuss fact and fiction regarding decision support systems beginning with information systems that support the decision making process to more sophisticated knowledge based systems that provide interpretation and analysis of data.

Computerized medical records--new opportunities.

The new generation of computer hardware and software is providing many new opportunities for physicians to use computerized medical records in their practices. A substantial number of new commercial systems have been introduced within the past few years, and many more will likely become available soon. In addition, there are opportunities for physicians to create their own customized systems using new software tools now available.

Collecting, storing and transmitting heart sounds.

We have developed a system whereby heart sounds and accompanying electrocardiogram may be recorded in digital format, stored as files and transmitted as any other digital file. The present system is implemented on a Macintosh System 7.1 platform, but will run under System 7.0 to System 7.5. The code is written in object oriented C++. The data are presently taken using a standard heart sound microphone and ECG amplifier. The data collection program stores the sampled channels as 16 bit words. Each channel is sampled at 2 kb/s for periods of time ranging from 10 to 30 seconds. Our program then reads the file, displays it under a viewer and allows for high fidelity reproduction of the sounds.

Cost-effective clinical uses of wide-area networks: electronic mail as telemedicine.

Electronic mail (E-mail) is widely used as a means of communication in the medical community. E-mail is clearly inexpensive when compared to two-way, fully interactive, real-time, video telemedicine. By content analysis of 200 consecutive messages, we show E-mail to be a low-cost use of computer networks, supporting a wide range of physician decision-making.

Comparing clinical vocabularies using coding system fidelity.

Much effort has been directed toward the development of an ideal multipurpose controlled medical vocabulary for use in human and veterinary medicine. SNOMED International is one effort that has resulted in a larger and more complex nomenclature system. Although it was able to code more concepts, SNOMED International failed to statistically improve vocabulary fidelity when compared with the 30+ year old SNVDO vocabulary. We found that SNOMED has a lower intercoder consistency than SNVDO and that a greater number of codes were necessary to represent an individual concept. Our study shows a significant Coder-Vocabulary interaction which suggests that more emphasis should be placed on coding guidelines and coder training. Clinician data entry and coding may be necessary for maximum vocabulary fidelity.

Integrating molecular diagnostic and flow cytometric reporting for improved longitudinal monitoring of HIV patients.

Studies have shown that monitoring HIV-infected patients undergoing antiretroviral therapy is best represented by combined measurement of plasma HIV-1 RNA and CD4+ T-lymphocytes [1]. This pilot study at the University of Missouri-Columbia integrates molecular diagnostic and flow cytometric data reporting to provide current and historical HIV-1 RNA levels and CD4+ T-cell counts. The development of a single database for storage and retrieval of these values facilitates composite report generation that includes longitudinal HIV-1 RNA levels and CD4+ T-cell counts for all patients. Results are displayed in tables and plotted graphically within a web browser. This method of data presentation converts individual data points to more useful medical information and could provide clinicians with decision support for improved monitoring of HIV patients undergoing antiretroviral therapy.

Design of access control methods for protecting the confidentiality of patient information in networked systems.

Public awareness of the potential for violation of personal privacy in clinical information systems is increasing. Much of this increase can be attributed to the popularity and publicity of the World Wide Web. Nightly news reports of intruder break-ins and flaws in Internet software security have stimulated public interest in the security of clinical information systems available over the web. As part of the development of systems designed to provide clinical narratives to physicians over the Internet, we are exploring designs that provide additional protection and security to these systems. Specifically, we are developing and testing automated access control measures based on provider-patient relationships for controlling access to personally identifiable patient information.

Providing location-independent access to patient clinical narratives using Web browsers and a tiered server approach.

Health care today depends upon timely access to patient medical data and the latest medical knowledge. As we make the transition from a hospital-based organization to an integrated health care delivery system, patient care information must move throughout the organization quickly and efficiently over increasing distances. The emergence of widely-dispersed referral networks demands novel solutions to the problems of delivering patient care information to providers. We have developed a mechanism to provide location-independent access to clinical narrative reports using a multi-tiered server model and World Wide Web technologies for delivery. To successfully deploy such a system to sites separated by large distances, it is important to reduce complexity at the client site. Using a "thin client", such as a web browser, in our design facilitates deployment and support while reducing cost per user. This architecture allows the application to be updated without modification to the end-user software and eases maintenance over long distances.

A system for browsing the SNOMED International vocabulary.

We have developed a means for "browsing" an extremely large vocabulary of over 130,000 medical terms (SNOMED International, 3rd ed.). This vocabulary someday hopes to become a standard vocabulary and coding scheme for patient medical records. After obtaining the vocabulary in machine readable form (ASCII in standard data format--SDF) we wrote a meta program in Turbo Pascal which parsed and embedded control codes into an input file for importing the data into an "infobase" known as FolioVIEWS. Using this infobase, it became possible to rapidly and reliably navigate around the SNOMED vocabulary in a manner impossible with printed material. The FolioVIEWS search engine is robust and rapid and allows for easy access to the terms. The utility of the infobase in evaluating the SNOMED vocabulary has been quite good. The actual utility of SNOMED as a standard vocabulary remains to be evaluated.

Delivery of laboratory data with World Wide Web technology.

We have developed an experimental World Wide Web (WWW) based system to deliver laboratory results to clinicians in our Veterinary Medical Teaching Hospital. Laboratory results are generated by the clinical pathology section of our Veterinary Medical Diagnostic Laboratory and stored in a legacy information system. This system does not interface directly to the hospital information system, and it cannot be accessed directly by clinicians. Our "meta" system first parses routine print reports and then instantiates the data into a modern, open-architecture relational database using a data model constructed with currently accepted international standards for data representation and communication. The system does not affect either of the existing legacy systems. Location-independent delivery of patient data is via a secure WWW based system which maximizes usability and allows "value-added" graphic representations. The data can be viewed with any web browser. Future extensibility and intra- and inter-institutional compatibility served as key design criteria. The system is in the process of being evaluated using accepted methods of assessment of information technologies.