5
th
International Advanced Technologies Symposium (IATS’09), May 13-15, 2009, Karabuk, Turkey
© IATS’09, Karabük University, Karabük, Turkey
RECOGNITION OF DENGUE DISEASE PATTERNS USING ARTIFICIAL
NEURAL NETWORKS
B. Gultekin Cetiner
a
, Murat Sari
b
and Hani M. Aburas
c
a, *
Faculty of Engineering, Department of Manufacturing and Materials Engineering,
IIUM, Kuala Lumpur 50728, Malaysia. E-mail: drcetiner@iiu.edu.my
b
Pamukkale University, Faculty of Art and Science, Department of Mathematics, 20020, Denizli, Turkey
E-mail: msari@pau.edu.tr
c
Faculty of Engineering, Department of Industrial Engineering,
KAU, P.O. Box 80204, Jeddah 21589, SA E-mail: haburas@kau.edu.sa
Abstract
This research aimed at the recognition of the patterns for
dengue disease patterns using Artificial Neural Networks
(ANN’s). Real data was provided by Singaporean National
Environment Agency (NEA), for academic purposes only.
Obtained data was used to model the behavior of dengue
cases based on the physical parameters of mean
temperature, mean relative humidity and total rainfall. The
set of data recorded weekly consists of dengue reported
confirmed cases together with three aforementioned
parameters and for a six-year period, January 2001 to April
2007.
Keywords: Disease Pattern Recognition, Artificial Neural
Networks
1. Introduction
Dengue is an infectious disease transmitted from person to
person by a mosquito, Aedes aegypti, which is a major
vector for the virus in different parts of the globe; Ae.
albopictus is also considered as a secondary vector, Fig 1.
The first recorded epidemics of dengue like disease
occurred in 1635 in French West Indies, Batavia (Jakarta),
and Cairo [1, 2, 3]. Dengue out-breaks have also been
reported from different parts of the world in the past two
centuries. Over the past two decades, there has been a
dramatic increase in the Dengue Hemorrhagic Fever
(DHF) and Dengue Shock Syndrome (DSS) epidemics in
South East Asian countries. World Health Organization
(WHO) estimated that about 50-100 million cases of
dengue are recorded from all over the world annually, and
two fifth of the world population is at risk and more than
one hundred countries have been affected by dengue or
DHF/DSS epidemics. Since 1950, more than 500,000
hospitalized cases and approximately 70,000 deaths of the
children have been recorded; infection rate among the
children is as high as 64 per 1000 population [2].
Dengue is characterized by high fever, headache, pain in
various parts of the body, prostration, rash
lymphodenopathy, and leucopenia [1]-[4]. DHF is a severe
febrile disease characterized by abnormalities of
homeostasis and increased vascular permeability which
may result into DSS [5]-[6]. Normally, cycles of dengue
virus is transmitted from human-to-human by mosquito
bites. Several species of mammals and lower primates act
as reservoir of the dengue virus [7]. From feeding on an
infected viraemic human, the female Aedes mosquito is
able to transmit dengue virus after an extrinsic incubation
period of 8-10 days [8]. Aedes mosquito rests inaccessible
areas behind the human dwellings; hence, the collection of
these mosquitoes by hand catch is very difficult. However,
the adult mosquitoes are being collected by either man-
biting/landing or netting. These methods are considered
as unethical issues for measuring the adult population.
Therefore, attempts are being made to collect these
mosquitoes through different types of traps developed by
different companies. In this study, various types of traps
have been tested to choose the most efficient one for the
collection of Aedes population. As per our study, Black
Hole traps was considered as the most efficient traps,
which was used in different parts of the study areas.
Figure 1: Dengue Transmitting Vector World Wide
(Source:www.broad.mit.edu/news/links/dengue-
08172005.html , 2005)
The main purpose of this study is to investigate the
feasibility of applying the Artificial Neural Networks (ANN’s)
technique to recognize dengue disease patterns based on
the measured real parameters: mean temperature, mean
relative humidity, total rainfall and reported dengue cases
as corresponding output to those three mentioned
parameters.
ANN’s have been used herein due to their ability to learn
from given examples which makes them perfect tools since
there is no need to model each individual case
mathematically.
In this paper, we use Artificial Neural Networks to
recognize the number of dengue-confirmed cases. The
real data obtained from the Singaporean National
Environment Agency (NEA) has been adopted to model
the behavior of dengue confirmed cases based on the
parameters of mean temperature, mean relative humidity
and total rainfall. The set of data recorded weekly consists