Decision Support Systems
Computer systems that provide users with support to analyze complex information and help to make decisions are called decision support systems (DSSs). In some cases, such systems may predict the impact of decisions before they are made.
Decision support systems are an advancement of management information systems, whose main purpose is to supply information to managers. Although the functions of management information systems are limited mostly to control of storage, organization, retrieval, and maintaining the security and integrity of data, decision support systems include model building and model-based reasoning capabilities. Decision support systems do not have the problem-solving competence of expert systems, however, which are programmed with the knowledge of human experts and can make some final decisions without human intervention. Decision support systems generally help human beings solve complex problems, and provide data that can lead to non-predetermined solutions that are beyond the limitations of expert systems.
The concept of decision support systems originated in the late 1950s and early 1960s, when researchers at Carnegie Institute of Technology located in Pittsburgh, Pennsylvania, conducted pioneering studies of organizational decision-making. Some time later Tom Gerrity and a team of scientists performed research on interactive computing at the Massachusetts Institute of Technology. One of their projects was dedicated to supporting investment managers in the administration of their clients' stock portfolios. These original studies were mostly done by graduate students and professors at leading engineering and business schools who had access to the most advanced computer systems of that time. It became obvious at the time that the emerging computer technology could serve as an excellent tool for the development of decision support systems.
Later on, scientists and computer programmers applied analytical and scientific methods for the development of more sophisticated DSSs. They used mathematical models and algorithms from such fields of study as artificial intelligence (AI), mathematical simulation and optimization, and concepts of mathematical logic.
In recent years, decision support tools like data warehousing and online analytical applications have enhanced the capabilities of decision support
systems. Joint efforts of scientists and programmers resulted in the development of the DSS generator—software for the building of decision support systems customized to the user's requirements.
Applications
Over the years DSSs have gained popularity in such areas as business management, medicine and health care, the military, environmental policy, and other areas that involve risk management, proper resource allocation, and similar tactical and strategic decisions. Decision-making is particularly important in the areas of business and finance. Very often, distribution of valuable resources and large sums of money are based on human intuitive judgment. DSSs provide aid in such tasks as cash flow analysis, break-even analysis, scenario analysis, and inventory techniques. The number of working DSSs demonstrates the necessity of such systems. The Taxpayers' Assistant System of the U.S. Internal Revenue Service helps its personnel give correct tax information to taxpayers. Also, DSS programs assist credit card company employees in allowing or disallowing certain charges, and advise bank officers regarding loans and mortgage approvals.
In manufacturing, decision support systems are solving design problems by using analytical, statistical, and model-building software, especially in the car- and aircraft-making industries. Such giants of automotive industry as General Motors and Ford Motor Company use comprehensive decision support systems.
Systems for scheduling and customer support are universally available for use in different fields. Their use helps bring down costs and increase service efficiency. American Airlines developed a system to schedule airplanes for required maintenance. As people deal with increasingly complex machinery and sophisticated instrumentation, DSSs become an important component in service and customer support, particularly providing expertise in troubleshooting and systems operation.
Decision support systems are also widely used in environmental science, particularly in air quality impact analysis. For example, a model developed by the Stanford Research Institute computes hourly averages of carbon monoxide for any urban location, while a program developed by the Argonne National Laboratory of Illinois provides meteorological data and classifies sources of emission for the Federal Aviation Administration (FAA).
Health-care professionals are assisted by decision support tools in determining the prognosis of individual patients based on an analysis of clinical data, and in determining whether a patient is eligible for clinical research based on the patient's clinical history. Applications such as alerting systems, critiquing systems, and diagnostic suggestion systems are utilized in hospital operating rooms, intensive care units, laboratories, and other medical facilities. Implementation of these tools can significantly improve the quality and reduce the cost of health care.
These and other DSSs may work actively or in a passive mode. Passive systems are mostly used by physicians for reference purposes, while active systems give advice in certain situations, such as alerting medical personnel when a parameter being monitored in a patient—such as temperature or heart rate—exceeds its designated threshold value.
Whether they are active or passive systems, health care decision support systems are limited to the role of advisor to clinicians and other health care professionals. They do not act as expert systems. Only physicians themselves are responsible for decisions made concerning human health; DSSs act as information tools in this and other circumstances.
Components
An effective decision support system requires reliable data and usually includes an interactive user interface, a knowledge base, and an inference (reasoning) engine.
The knowledge base is the component of a DSS that contains both well-established facts and results obtained by trial-and-error methods. It may include such different elements as formulas (algebraic, logical, or statistical expressions), guidelines, and knowledge of risk and cost of operations. Often knowledge is presented in the form of rules.
The inference engine is a complex computer algorithm, which uses the data in the knowledge base to obtain a solution to a problem. There are two approaches for decision-making systems: under certainty and under uncertainty. Decision-making under certainty most often utilizes a mathematical model that gives an unambiguous response. On the other hand, decision-making under uncertainty uses statistical approaches, such as Bayesian networks, neural networks, and other methods of AI.
The advancement of decision support systems would be impossible without its interaction with a number of related fields. Business analysts and mathematicians develop mathematical models for use in DSS, software engineers provide tools for knowledge maintenance and data mining, and psychologists assist in DSS design by conducting behavioral decision-making research.
As the need for more sophisticated decision-making will increase in the future, it is probable that decision support systems will become more intelligent and user friendly, and applicable to a broader spectrum of professions.
Marina Krol and Igor Tarnopolsky
Bibliography
Berner, Eta S., ed. Clinical Decision Support Systems: Theory and Practice. New York: Springler-Verlag, 1999.
Krol, M., and D. Reich. "Development of a Decision Support System for Detecting Critical Conditions During Anesthesia." Journal of Medical Systems 24, no. 3 (2000): 141–146.
Marakas, George M. Decision Support Systems in the 21st Century. Upper Saddle River, NJ: Prentice Hall, 1999.
Turban, Efraim, and Jay E. Aronson. Decision Support Systems and Intelligent Systems, 6th ed. Upper Saddle River, NJ: Prentice Hall, 2001
