Prediction of Infectious Fevers Using Big Data Analytics

To Prevent Future Outbreak

Introduction:
Today is the era of Big Data. The need to analyse vast amounts of data, being generated in
different fields ranging from medicine to financial markets and from transportation to
environmental modelling, is emerging as the next big challenge and opportunity.
In order to create awareness about developments in the emerging field of Big Data -- including
for example new algorithms and systems design for Big Data and new concepts in machine
learning related to Big Data -- and to bring together the broader community so as to leverage
existing strengths, we are initiating a series of public lectures on various aspects of Big Data,
open to both academia and industry, as a platform for open exchange of ideas.
Searches Medical databases for hidden patterns, finding predictive information that experts may
miss because it lies outside their expectations. At present dengue fever is positive only after one
week. In our research data mining technique is used to predict the useful information to predict
the type of fever. Our research mainly focus on predicting the dengue fever in the day one based
on the symptoms, age and laboratory reports by Big Data Analytics (Hadoop)
In future this application can be extended to hospitals to predict the type of fever in
outpatients.
Origin of the research problem :
Dengue fever is an infectious disease caused by virus and transmitted to persons by mosquitoes.
A person contracted dengue fever usually shows symptoms including sudden spike of high fever,
muscle ache, joint pain, rashes, and headache. An infected person normally recovers within a few
weeks, but a small per cent of patients can develop fatal complications such as plasma leakage,
severe bleeding, and severe organ impairments which can lead to death. To date, there is no drug
for treatment or vaccines to protect against dengue fever. Therefore, the most effective way to
prevent dengue is to control the mosquito population [1].
Health accounts for significant mortality in developing countries. Dengue fever serology takes a
week at least to be reported positive signs, thus necessitating the need for other markers of
diagnosis and prognosis [2]. A medical database contains information related to patients. This
data may go wasted if we have not used efficiently. The Big data analytics can be applied on this

data of databases to predict the patients type of fever (Dengue/malaria/swine flu /typhoid/other
viral fever etc).
Interdisciplinary relevance:
The Big Data subject is an interdisciplinary relevance. The proposed work is of
interdisciplinary in nature. The Big Data Analytics are applied on medical database
to predict the disease in early stages. It would be more useful for diagnostic
laboratories, Biochemistry labs, Pharmaceutics etc.

Review of Research and Development in the Subject:

It is pursued the existing literature briefly and confirmed that there is no findings came out from
the Computer Scientists from this Medical data to find the dengue fever in the early stages.
National Status: Doctors understand the weakness of the present situation and create the
premises for innovations by filling in the gap between present concept and the future
preferences. In Medical labs dengue fever is tested positive only after one week.
Diagnosis of dengue fever(DF) can be done by identifying the influential clinical
symptoms and laboratory features without a clinical expert[1]. A computational
intelligence-based approach that predicts the diagnosis of dengue fever in real time, is
proposed.
International Status: Warning System Predicts Outbreaks of Dengue Fever [3] Apr. 29, 2013With the help of a warning system which measures the risk of dengue incidence using
precipitation and air temperature, it is possible to forecast the outbreak of dengue fever up to 16
weeks in advance. But this is only predicts environmental conditions that due to mosquitoes
human may get dengue fever. There is no such expert system that predicts the dangerous dengue
fever early in the human body. Health accounts for significant mortality in developing countries.
A novel dengue fever (DF) and dengue hemorrhagic fever (DHF) analysis using artificial neural
network (ANN)[3]. The developed system bases its prediction solely on the clinical symptoms
and signs and uses the multilayer feed-forward neural networks (MFNN). The results show that
the proposed system is able to predict the day of defervescence in dengue patients with 90%
prediction accuracy[4].

Significance of the study:

Pattern discovery through Big Data Analytics such as characterization, discrimination,
association, classification, clustering, outlier and trend analysis, etc. searches databases for
hidden patterns, finding predictive information that experts may miss because it lies outside
their expectations. This information is very much useful to the doctors to give suitable
treatment.
This expert system can be used to predict not only dengue fever but it can also be used to
predict most dangerous diseases like malaria, swineflu etc.,
Objectives:
(a) The purpose of this Research project is to investigate the role of data mining technique as
a tool in diagnosing and predicting the severity of dengue fever in the early stages.
(b) Collect the data from medical databases and preparing the training set.
(c) Developing an efficient model to predict the hidden information from medical databases
(d) Applying the training set on the model to get the required result. This process continues
until we get accuracy.
(e) Testing the model by using teat data.
(f) Finally apply the new data that is patient data to predict the type of fever.
Methodology:
(1) Collect the medical data from various highly standard labs and hospitals:
Medical data related to all types of fever is collected from Hyderabad, Bangalore, Madras
and Delhi.
(2) Analyze the data and prepare the training set (known data)
Collected data is separated based on the type of fever. Here major diseases related to
fever like Dengue, Malaria and swine flu are separated. This is called training set.
(3) Construct expert system based on the training set using Big data Analytics

(4) The expert system is repeatedly tested:

The training set is applied as input to the expert system by using Back propagation
technique. Expert system is tested for accuracy by using test data.
(5) Expert system is used to predict the information:
The data of new patient i.e symptoms is applied to the expert system to predict the type of
fever. The type of fever can be classified based on the output.

References:
1. Revathi N. Prof.S.J.K.Jagadeesh Kumar. International Journal of Computer Applications &
Information Technology Vol. II, Issue I, January 2013 (ISSN: 2278-7720).

2. K. R. Bharath Kumar Reddy, Ramesh R. Laksmana, Basavaraja G. Veerappa, Shivananda

Ultrasonography as a tool in predicting the severity of dengue fever in childrena useful
aid in a developing country.Published online 3rd March-2013 @ springer-verlag Berlin
hedelberg-2013
3. http://www.sciencedaily.com/releases/2013/04/130429095052.htm.

Yien

Ling

Hii

concludes in the dissertation she is defending at Ume University in Sweden on 3 May.

4. Fatimah Ibrahim,mohdnaasirtaib, wan Abu bakar wan Abas,. A novel dengue fever (DF) and
denguehaemorrhagic fever (DHF) analysis using artificialneural network (ANN). Computer
Methods and Programs in Biomedicine (2005) 79, 273281.Elsevier