Information Technology Reference
In-Depth Information
In our approach we use a methodology of fuzzy logic, as the elements of the diag-
nostic process (symptoms, diagnoses, connections between both) and a diagnostic
process itself are mostly described imprecisely and approximately. Moreover, a
“normal” disease (with typical clinical picture/course) is not a unique (crisp) con-
cept. A “normal” disease can be also categorized as, for example, confirmed, pos-
sible, etc. Thus, we should define deviations from non-uniquely defined “normal”
diseases that can be interpreted as a marker of the RD.
In our work we restrict ourselves to a limited number of child/adolescent dis-
eases, particularly, pneumonia, bronchitis, atopic dermatitis, some others.
We obtain a description of these diseases from the experts (manuals): preliminary
information about “normal” diseases is summarized in linguistic and numerical form
(the details will be given below). This description is done based on the following
criteria:
a) general properties that include, for example, the following parameters: gender,
age, etc.
b) interior properties, that include how a patient himself describes his/her state.
The problem is that in pediatrics it is difficult to obtain such description from
children younger than 5
7 years old.
c) exterior properties are obtained from
a) persons who observed a child (parents, relatives, etc.)
b) physician's observation.
c) lab and instrumental tests.
d) observation of a disease duration and reaction on a treatment.
Subjectivity is present in most of these criteria.
25.3
Preliminaries
25.3.1
Notions and Denotations
As was already discussed [12-14], four components - symptom-disease relation,
patient information, patient-disease relation and an inference mechanism with cor-
responding denotation
- should be defined to describe a medical
decision-support system within the fuzzy logic framework.
Let the knowledge base of the proposed decision support system consists of fuzzy
R SD ,
S p ,
D p ,◦
IF
THEN rules that describe the relationships between symptoms/signs, test re-
sults and findings - for all these medical entities we use a term "symptom/sign" or
its abbreviation " S "; and diseases, diagnoses - with denotation " D ". Thus, our rules
look like “IF S THEN D ”. Fuzzy relations between symptoms/signs and diseases
can be defined as
R SD :
Σ × Δ [
0
,
1
]
(25.1)
To build these fuzzy relations the crisp sets of patients
Π = {
p 1 ,...,
p r }
, symp-
toms/signs
Σ = {
s 1 ,...,
s m }
and diseases
Δ = {
d 1 ,...,
d n }
under consideration are
Search WWH ::




Custom Search