Knowledge Discovery and Data Mining in Medicine - Advanced Techniques in Knowledge Discovery and Data Mining - page 202

H+K threshold values; we can set ( H ( M+K ) +H+K ) as F in Eq. (7.29). Therefore,
Eq. (7.29) becomes Eq. (7.32).

AIC = − l + H M + K + H + K (7.32)

2 ) 2 ( ) ).

Fitness Function with AIC
To include not only error estimation but also the goodness of the network structure,
we define the fitness function with AIC as follows:
()

λ − AIC

max u

Fitness u = , (7.33)

λ − λ

max min

where is the minimum value of

AIC. The u is the index of an individual in the population.

λ is the maximum value of AIC and λ

max min

7.4.4 Immune Cells by PLNNs

Macrophage employs the PLNN to classify the training cases. The hidden neurons
are generated/annihilated during the learning phase, and consequently the
remaining neurons are assigned to the corresponding subset of training cases,
respectively. T-cell employs neural network learning to assign a training case into
one of the B-cells. In this chapter, T-cell employs the lower part of PLNN and the
network enforces learning reverse signals from output neurons as shown in Fig.
7.11. Because T-cell also recognizes input signals, T-cell trains the lower part of
PLNN simultaneously. The teaching signals in the network consist of binary
strings: 1 (on) and 0 (off). In biological immune systems, B-cells receive
stimulation from T-cells. In our model, B-cells employ a simple DNN learning
method [1] to train a network for the subset of training cases assigned by T-cell, as
shown in Fig. 7.18. Although B-cells work to learn a subset of training cases
independently, B-cells cooperate with each other in a classification task, as shown
in Fig. 7.19.
Figure 7.20 shows the reasoning process of the trained IMANN. After
training PLNN, an arbitrary input is given to T-cell NN. T-cell NN classifies into a
group and stimulates the corresponding B-cell NN. The B-cell NN calculates
output activities as a total output of IMANN.

7.4.5 ICU Database

To demonstrate the effectiveness of our scheme, we use the intensive care unit
(ICU) database [23], [24]. The variables incorporated into the models were gender,
age (<44, 45-64, 65-74, 75+), severity of illness (0-9, 10-19, 20-29, 30-39,
40-49, 50-59, 60+; predictive hospital mortality rate derived from APACHE II
score [21]), operation (none, elective, urgent), ventilator (yes, no), urinary catheter
(yes, no), central venous catheter (yes, no), and infection (none, drug-susceptive,
drug-resistant). The signal of an output neuron was “0” or “1” representing “dead”
or “alive.” Table 7.12 shows the variables in the ICU database. Figure 7.21 shows
relationships between the variables in the ICU database.