Information Technology Reference
In-Depth Information
of a particular medical hypothesis. For patient administration, data are collected
and used for accounting and planning [6]. As a result, data mining uses heteroge-
neous data and multiple meanings. Integration of these diverse data and metadata
requires creation of a unified conceptual data model. This task is complicated by the
fact that data are collected using diverse collecting methods, inclusion criteria, sam-
pling methods, sample sizes, types and numbers of measurements, and definitions
of outcomes. Therefore, conceptual data modeling for data mining must address the
problems related to the secondary use of the data, which means that the data were
originally collected for other purposes and may have more or less explicitly defined
meaning (metadata).
In his
Handbook of Analytic Philosophy of Medicine
, Dr. Sadegh-Zadeh outlines
four main classes of medical concepts: individual, qualitative (classificatory), com-
parative, and quantitative. Individual concepts denote specific individual objects.
For example, John Davey denotes specific patient. Qualitative or classificatory con-
cepts are “either unary predicates or non-comparative, many-place predicates.” For
example, John Davey has excessive daytime sleepiness, EDS. Comparative con-
cepts express relationship between two or more objects. For example, John Davey
has higher EDS than John Smith. Quantitative concepts specify the magnitude of
an attribute using a numerical function. For example, John Davey's sleepiness can
be “measured” using the Epworth Sleepiness Scale (the standard questionnaire used
in sleep clinics). Thus, we can say that John Davey has sleepiness of 20, and John
Smith has sleepiness of 15. Dr. Sadegh-Zadeh defines quantitative concepts as “a
homomorphism
f
from a particular empirical, i.e., experiential, structure
into
a numerical structure
R
such that
R
is, usually, the set of positive real numbers, and
≥
R
,≥
is the 'is greater than or equals' relation.” The quantification f is also called mea-
surement and the homomorphism is referred to as a scale. Quantitative concepts
are particularly important in medical diagnosis and treatment evaluation. We will
discuss them further in context of the operational definition of concept.
Additionally, Dr. Sadegh-Zadeh discusses the notion of an attribute, and de-
scribes two types of attributes: categorical and dispositional. Categorical attributes
are permanently present, for example, John Smith's weight. The specific value for
weight may change with time, but weight is measurable under all circumstances.
On the other hand, dispositional attributes may manifest only under certain circum-
stances, for example, John Smith's daytime sleepiness. Daytime sleepiness can be
observed during day (assuming that John Smith is not a nightshift worker).
13.3.1
Operational Definition of a Concept
Operational definition is based on the notions of conceptualization and operational-
ization. Conceptualization defines the concept in the context of its use, for example,
excessive sleepiness in the context of the clinical diagnosis of a specific sleep dis-
order. Operationalization states that a concept is defined in terms of the operations
by which its referent is measured. For example, sleepiness is “measured” by a sub-
jective measure [16]. The Epworth Sleepiness Scale (ESS) is a self-administered
