5.2 Proposed Diagnostic Model
This section describes the proposed diagnostic model for the early
detection and prediction of dengue disease. A PSO-ANN based classifier
is applied to accurate detection of dengue disease. In the proposed
classifier, PSO is applied to optimize the weight and bias parameters of
ANN technique. In the proposed model, 10-cross fold validation technique
is implemented which divides the dengue data into ten equal parts. The
detailed description of this technique is given in subsection. In the
proposed model, sixteen input neurons are defined in input layer
corresponding to sixteen attributes of dengue dataset. The hidden layer
consists of seventeen neurons, whereas output layer consist of only two
neurons corresponding of class labels such as dengue positive and dengue
negative. The weight of each neuron is computed using PSO technique and
the optimized weight is used to train the neural network. The
classification accuracy and confusion matrix are considered as a
performance parameter to evaluate the performance of the proposed
diagnostic model. Fig. 4 shows the proposed diagnostic model for earlier
detection of dengue disease.