Information Technology Reference
In-Depth Information
and CDSS systems, and discuss the similarities and differences between these three
terms. Specifically, we focus on the difference between ES and DSS systems in
terms of the users' requirements.
Expert System (ES) can be defined as a software system which uses knowledge
and intelligent reasoning for complex problem solving in, often, highly specialised
areas of expertise. ES systems are often viewed as a subject of the AI research
[1, 12, 15]. ESs have four characteristics which differentiate them from other infor-
mation systems [15]: (1) explicit representation of knowledge; (2) ability to explain
problem solution (how explanations) and ability to give explicit reasons for using
particular solution (why explanations); (3) utilization of logical reasoning, as op-
posed to mostly imperative algorithmic approach used in information systems; and
(4) processing based primarily on a symbolic approach.
Decision support system (DSS) has been defined in different ways by many
ES and DSS researchers [5, 10]. In our research, we use a broad DSS definition
given by Finlay [5], who describes DSS as “a computer-based system that aids the
process of decision making”. We view DSS as a wide-ranging category, which
includes various subtypes identified by Power [10]: data-driven, model-driven,
communication-driven, document-driven and knowledge-driven. We focus here on
the knowledge-driven DSS systems, which use explicit knowledge representation
and, therefore, are closely associated with the construction of ES systems.
Clinical Decision Support System (CDSS) can be defined as a specialized know-
ledge-based DSS used in medicine. Berner and La Lande [2] define CDSS as “a
computer system designed to impact clinician decision making about individual pa-
tients at the point in time that these decisions are made.” In general, the knowledge-
based CDSS have three main components: the knowledge base (KB), the inference
or reasoning engine, and the input/output mechanism to communicate with the users
[2]. CDSS have been incorporated in medical systems for a long time, and they vary
in the type of support they may offer. For example, CDSS differ in whether they are
stand-alone or embedded in other systems such as, for example, electronic medical
records (EMR). Also, CDSS differ in whether they provide general or speciality-
based information and which clinical tasks they support. For example, many
successful CDSS systems support laboratory test ordering [11] and provide comput-
erized physician order entry (CPOE), which alert the users to possible interactions
between prescribed medications [8]. One of the well-known classes of CDSS is
clinical Diagnostic Decision Support System (DDSS). DDSS are defined by Miller
and Geissbuhler [8] as “a computer-based algorithm that assists a clinician with one
or more component steps of the diagnostic process.”
Most CDDS arose out of earlier expert systems research; however, the intent of
CDDS is not to simulate an expert's decision making, but to assist the clinicians
in their decision making process. As stated by Berner and La Lande [2], “the role
of the computer should be to enhance and support the human who is ultimately
responsible for the clinical decision.” The users of the CDSS actively interact with
the system. They expect the CDDS system to provide deep explanation of suggested
decisions. The users of CDDS want to be able to follow the reasoning, filter the
information, discard useless information, utilize the useful information, and make
