Biology Reference
In-Depth Information
1
5
ð
11
:
6
10
þ
1
:
6
Þ¼
¼
2
:
32
:
5
This will be the low BG index for these five readings.
Similarly, the high BG index is:
1
5
ð
1
5
ð
112
:
5
2
2
2
10
ð
2
Þ
þ
10
ð
1
Þ
þ
10
ð
2
:
5
Þ
Þ¼
40
þ
10
þ
6
:
25
Þ¼
¼
22
:
5
:
5
This example was an illustration for the following general situation. Let
x
1
, x
2
,
x
n
be n BG readings of a subject and let:
...
rl
ð
BG
Þ¼
r
ð
BG
Þ
if f
ð
BG
Þ
< 0 and 0 otherwise
;
(5-5)
rh
ð
BG
Þ¼
r
ð
BG
Þ
if f
ð
BG
Þ
< 0 and 0 otherwise
:
The LBGI and the HBGI are then defined as:
n
X
n
1
LBGI
¼
rl
ð
x
i
Þ
i
¼
1
n
X
n
1
HBGI
¼
rh
ð
x
i
Þ:
i
¼
1
The LBGI is based on the left branch of the BG risk function, whereas
the HBGI is based on the right branch of the BG risk function
(see Figure 5-8).
E
XERCISE
5-3
Give factors that will cause the LBGI and HBGI to increase.
VIII. MODEL VALIDATION STRATEGIES
In the preceding sections, we developed a mathematical model of a
quantitative risk measure that, we claim, holds promise in assessing the
clinical risk for BG deviations from the safe target range. As with any
new mathematical model, however, the burden of proof for its validity
and usefulness lies with its creators, and in this section we address the
validation question.
To assess the performance of our model, we need to test it on data. The
model uses assumptions on SMBG readings; thus, the data we need
must be of this form. Several questions arise.
First, how are these data obtained? This question is logistic in nature—
when humans are used as test subjects, there are strict government
