Pattern Recognition Techniques - Biomedical Signal Analysis: Contemporary Methods and Applications - page 197

g(x) Se + Sp +

g normal (x)

(x)

g

Discriminant functio

abnormal

for normal class

Discriminant function

for abnormal class

x

TN FN FP

g (x)

abnormal

= 1,0

T at

g

normal (x)

Figure 6.15
Discriminant functions for two populations, one with a disease and the other
without the disease. A perfect separation between the two groups is rarely given; an
overlap is mostly observed. The FN, FP, TP, and TN areas are indicated.

determine the patients with disease, and the specificity shows the ability
of the test to determine the patients who do NOT have the disease.
In general, the sensitivity S e

and the specificity S p of a particular

test can be mathematically determined.
Sensitivity S e reveals that the test result will be positive when disease
is present (true positive rate, expressed as a percentage):

TP
FN + TP

S e = (6.49)

Specificity S p is the probability that a test result will be negative when
the disease is not present (true negative rate, expressed as a percentage):

TN
TN + FP

S p = (6.50)

Sensitivity and specificity are functions of each other and also counterre-
lated. The x -axis describes the specificity and the ROC curve expresses
1-specificity. Thus, the x and y coordinates are given as

TN
TN + FP

x =1 − (6.51)

TP
FN + TP

y = (6.52)