Complex Measurement Classification in Medical Applications Using a Case - Based Approach

TitleComplex Measurement Classification in Medical Applications Using a Case - Based Approach
Publication TypeClassical
Year of Publication2003
AuthorsNilsson M, Funk P, Sollenborn M
Number of Pages6
PublisherWorkshop proceedings of the fifth international conference on case-based reasoning
CityTrondheim, Norway
Abstract

In
many domains it is sufficient to classify measurements o
b
tained by
sensors by filtering noisy data and performing a classification of the measur
e-
ments by means of some mathematical function. Classification of measur
e
men
ts
is difficult in
some medical domains, e.g. determination of stress levels
.
Mea
s-
urements and their classification may be too complex for an algorithmic a
p-
proach. Matching may result in a number of classif
i
cation candidates, historic
data (previously cla
ssified measurements) may be sparse, and same measur
e-
ments may indicate several diagn
o
ses. We propose a Case
-
Based Reasoning
approach in which feature ve
c
tors are matched against cases in a case library,
and in which indeterministic or weak classifications
are validated by an exper
i-
enced physician, pointing out the features relevant in the classification of ps
y-
chophysiological dysfunctions. The physician’s classific
a
tions are stored in the
case library, continuously improving the system’s pe
r
formance as it
is being
used. If only a few examples of classified measurements are available, the e
x-
perienced physician initially points out which feature co
m
binations are used in
the classific
a
tion/diagnosis.

URLhttps://www.researchgate.net/profile/Peter_Funk/publication/228898351_Complex_measurement_classification_in_medical_applications_using_a_case-based_approach/links/0912f5110fb71d7f1b000000/Complex-measurement-classification-in-medical-applications-using-a-