Dividing
the absolute frequencies in the 2x2 table by the total number of
events
N = A + B + C + D,
we obtain a useful model (sens=sensitivity, spec=specificity):

Disease
present (D+) 
Disease
absent (D) 
Test
result positive (T+) 
P(D)
x sens 
[1
 P(D)] x (1  spec) 
Test
result negative (T) 
P(D)
x (1  sens) 
[1
 P(D)] x spec 

A
+ B (diseased subjects) 
C
+ D (healthy subjects) 
Now
we are ready to derive a famous formula for the conditional probability
P(D+/T+)
This
is the famous
Bayes'
formula.
