PUBLICATION

PANDEMIC DISEASE DETECTION AND PREVENTION SYSTEM USING MINING WITH GRAPH-BASED
APPROACH

J.Sasitha Burvin,
PG Scholar,Department of Computer Science Dr.K.Dhanalaxshmi,
and Engineering,
PSNA College of Engineering and Technology, professor,Department of Computer Science and
[email protected] Engineering,
PSNA College of Engineering and Technology,
[email protected]

Keywords: Association Rule Mining, Graph Based

ABSTRACT

General health examination is an integral part of INTRODUCTION

healthcare in many countries. Identifying the patients at risk is Pandemic is An endemic occurring worldwide or over a very
important for early warning and preventive intervention. Data wide area, crossing international, boundaries and typically
mining is a well known technique used by health organizations affecting a large number of people (http://www.asset-
for classification of diseases. The clinical documents scienceinsociety.eu/pandemic). Dengue is fast emerging
maintained are a pool of information regarding the infected pandemic-prone viral disease in many parts of the world.
patients. Dengue is fast emerging pandemic-prone viral Dengue is also known as bone breaking illness. It is divided
disease in many parts of the world. It should also be noted that into two types as type 1 and type 2 namely Dengue Fever (DF)
dengue can be co-morbidity with other disorders. It can also and Dengue Hemorrhagic Fever (DHF) by the World Health
be detected in the patients with chronic disease. Here by using Organization. Dengue Hemorrhagic Fever (DHF) is again
Fever based dataset to be load and finding the patients affected classified into DHF 1, DHF 2, DHF 3, and DHF 4. It causes
by dengue fever prediction by machine learning. Association abdominal pain, hemorrhage, circulatory collapse, acute
rule mining is useful for analyzing, predicting patient’s platelet deficiency (Vadrevu et al, 2012).
behaviors’. Decision Tree is widely used by many researchers The symptoms of dengue include bleeding, low levels of
in healthcare field. In the proposed model Decision Tree helps blood platelets, low blood pressure and metallic taste in
to predict the dengue cases earlier and reduce mortality rate mouth, headache, joint pain and rashes. It is tough to
and classify different activities of patients in more accurate differentiate dengue fever and dengue hemorrhagic fever. The
manner. The patients affected with dengue fever were divided disease transmission happens when Aides Egypt mosquito
into those who are affected with dengue fever, or dengue fever bites a healthy person; the virus enters into the body fluids of
with chronic disease warning signs. The proposed work Using that person. Then it starts reproducing inside the white blood
graph based decision tree, it is possible to detect the dengue cells and initiates the dengue virus cycle.
fever and based on the complication of the patient disease A special type of EHR is the Health
prioritize the patients so that they will get effective treatment Examination Records (HER) from annual general health
in timely and accurate manner. check-ups (kumar et al, 2016). This is a virus born disease
caused by breeding of Aedes mosquito. In this paper we are interconnected and within the network they worked
discuss Association Rule Mining approaches of information together in parallel in order to produce the output functions.
mining that have been utilized for dengue disease
prediction(Ling Chen,2016). [2] Mazaher Ghorbani and Masoud Abessi(2017)have
Data mining is a well-known technique used by mentioned approach for the temporal information contain
health organizations for classification of diseases such as time-stamping information that affects the results of knowlege
dengue, diabetes and cancer in bioinformatics research mining. Traditional techniques for finding frequent itemsets
(Nilesh et al,2015). A suspected case is a clinically compatible assume that datasets are static and the induced rules are
case of dengue like illness, dengue, or severe dengue with an relevant across the whole dataset. However, this is not the case
epidemiological linkage. A probable case is a clinically when data is temporal. It improves the potency of mining
compatible case of dengue-like illness, dengue, or severe frequent itemsets on temporal data. Since patterns can hold in
dengue with laboratory results indicative of probable either all or number of the intervals, propose a new algorithm
infection. to restrict time intervals, which is called frequent itemset
A confirmed case refers to a dengue case that was mining with time cubes. Main focus is developing an efficient
confirmed by the serological tests IgM capture enzyme-linked algorithm for this mining problem by extending the well-
immunosorbent assay (ELISA) with single positive IgM in the known a priori algorithm
lab. Most analysis reviewed in this paper used the suspected
and confirmed cases. Dengue outbreak tracked, monitored and [3]Rituja A.Bibave1, DB. L. Gunjal(2017) The elemental
predicted to help the local authorities for future prediction. challenge of learning a classification model for risk prediction
The model generated can then be used to find and predict the lies within the unlabeled information that constitutes the bulk
presence of dengue. In addition to this, several prescription of the collected dataset. Significantly, the unlabeled data
drugs used for the management of chronic disease act describes the participants in health examinations whose health
adversely in dengue patients causing further complications. conditions can vary greatly from healthy to very-ill. There is
So, the patients affected with dengue fever are divided into no ground truth for differentiating their states of health. In this
those who are affected with dengue fever, or dengue fever paper, tend a graph-based, semi-supervised learning algorithm
with chronic disease warning signs. called SHG-Health (Semi-supervised Heterogeneous Graph on
Health) for risk predictions to classify a progressively
RELATED WORKS developing situation with the majority of the data unlabeled.
An efficient iterative algorithm is designed and the proof of
[1] A. R. Pon Periasamy and S. Mohan(2017)have convergence is given.
discussed an approach for the healthcare policies,
constructing drug recommendation systems, developing [4]Uthpala Premarathne and Adulate Alabdulatif (2016)
health profiles of individuals Medical databases are presents the HER method pertaining to “all aspects of care”
terribly large that need computerized programs to find latent (such as genomic test results, diagnoses, medication,
trends which will facilititate in diagnosing and treatment. By laboratory test results, and imaging data). Cloud-based utility
using the traditional methods it becomes very difficult in order services offer additional benefits to EHR system. They’re
to extract the meaningful information from it. These neurons more cost effective, can be easier to manage and support
collaboration, with mobile technologies and devices to gather PROPOSED ARCHITECTURE FOR DENGUE FEVER
data. PREDICTION

[6] Purushottam Sharma (2016) introduced Data mining The prediction of dengue infection carried out using data
algorithms, when aptly used, are capable of improving the mining techniques such as Association Rule Mining and graph
quality of prediction, diagnosis and disease classification. The based decision tree. Association rule mining algorithm useful
main aim is to analyze data mining techniques needed for for analyzing, predicting patient’s behaviors. In the proposed
medical data mining especially to find out the locally frequent work, a graph based decision tree is proposed to predict the
diseases such as heart ailments, lung cancer, and breast cancer dengue fever .This work proceeds to discuss with predicting
and so on. To evaluate the data mining techniques for finding patient’s behaviors as whether the patients are affected with
locally frequent patterns in terms of accuracy, cost, dengue fever or not. In addition to this, several prescription
performance, and speed. The extracted patterns displayed the drugs used for the management of chronic disease act
attribute relationships in time domain which helps in accurate adversely in dengue patients causing further complications.
diagnosis. So, the patients affected with dengue fever are divided into
those who are affected with dengue fever, or dengue fever
[7] R. Naveen Kumar and M. Anand Kumar suggested a with chronic disease warning signs. Overall, the proposed
paper on Medical Data Mining Techniques for Health care system will provide a new direction of predicting dengue with
Systems Due to the sequence in the information technology, chronic disease through data mining.
the prevalence of the healthcare organizations conserves their
data electronically. Enormous progress in medical data leads
to be scarce in the mining of well-informed in series from the
mass data. There is a necessity for accomplished analysis tools
to resolve covered relatives and desire in data. Data mining
can represent new biomedical and healthcare details for
clinical preference.

[8] Ling Chen and Xue Li (2016) introduced Personal health

indexing and geriatric medical examination. The demerits to
this methodology is to optimize problems that find optimal of
labels as health score based on medical records that are
infrequent, incomplete and sparse. Evolution of health care
status of a person from cradle-to-grave is becoming possible.

[9]Thanushka et al (2015) introduced that Four association

rule set summarization techniques.High compression rate
Loading Dataset
doesn't lead to low redundancy.
In this work process load to dengue dataset used laboratory
test results which has age, fever, mylagia, platelet-counts,
wbc-count, Elisa Test result (IgM,IgG)types of dengue virus low blood pressure and metallic taste in mouth, headache,
to process joint pain and rashes .We have to extract the data based on
Table shows the attributes and values. The Dataset should be status and follow up patient. So here analyzing what are those
loaded ,whenever the process executed. The dengue with symptoms are affected by patients and levels of dengue to be
chronicle disease dataset is loaded and after data is partitioned analyzed. Finding critical patients and details will be
and then finding results. classified through machine learning algorithms used.
Pre-Processing Finding User Behavior
A preprocessor is a program that processes its input data to Finding user behavior is the medical reporting information
produce output that is used as input to another program .Pre- about loading dataset. Here finding frequent patterns or items
through data classification methodology. . Association rule
Attribute Possible Value
mining algorithm useful for analyzing, predicting patient’s
EPID Any –alpha –numeric-value
behaviors then checking every attributes based on user health
Fever Yes or no
analyzing and finding behavior as normal or abnormal cases.
Bleeding Yes or no
Behaviors finding is based on patient’s health status reports
Mylagia Yes or no
and its disease levels.
Joint Pain/mettalic taste Yes or no
Performance Analysis:
Platlet Values
Classifier Performance Classifier performance is usually
WBC count/heart rate Values
measured by accuracy, the percentage of correct predictions
processing as suppose dataset loading with after any data over the total number of predictions made. Similar to decision
placement is to be null or unstructured data or unwanted data tree post-pruning, association rules can be post-pruned to
is to be removed. The output is said to be a preprocessed form reduce the number of rules produced. Many ideas on post-
of the input data, which is often used by some subsequent pruning of decision trees were introduced by Quinlan [Qui]. T
programs like compilers. Preprocess is removing null values Many other measures are also used to understand the different
or unstructured data from loading into the certain datasets. In aspects of the generated model such as: sensitivity, specificity
this process includes the following data pre-processing which precision and recall [HPY00]. The error rate signifies the
are used to make the modeling data. number of wrong predictions over the total number of
predictions. The accuracy rate signifies the number of correct
Data Partitioning predictions over the total number of predictions. These
Data Partition is also similar to data clustering. Cluster is a measures are defined as follows.
group of objects that belongs to the same class. In other words, Precision = true positives/ true positives + false positives
similar objects are grouped in one cluster and dissimilar Recall = true positives/ true positives + false negatives
objects are grouped in another cluster. Here different set of Accuracy = number of correct classifications/ total number of
attributes based load the certain dataset using clustering our classifications made.
data based on some valid attributes. Cluster analysis itself is To understand true/false positives and
not one specific algorithm, but the general task to be solved. negatives, let us use an example from information retrieval. So
Partition has our data set is splited into some of small files. So precision is a ratio of relevant results to all results and recall is
data set is partitioned for data classification method. Dengue is a ratio of relevant results to all relevant information.
also known as bone breaking illness. The symptoms of
dengue include bleeding, low levels of blood platelets,
Recall and precision are fused to form a single measure called
Emeasure.
Error = number of incorrect classifications/ total number of REFERNCES
classifications made 1. A. R. Pon Periasamy and S. Mohan (2017),’A Review
on Health Data Using Data Mining Techniques’,In:
Table shows the properties of these datasets and tested the set- International Journal of Advanced Research in
valued based classification system using a dengue dataset Computer Science and Software Engineering Volume
constructed by records. The illustration of the flow diagram 7, Issue 3.
for dengue prediction in Figure: 3.2 as above. This dengue
dataset consist of some instances. For experimental purposes, 2. Bui, N., Yen, J., & Honavar (2016). V,’Temporal
we used a 66% split, where 66% of the instances were used for Causality Analysis of Sentiment Change in a Cancer
training and the rest of the instances were used for testing the Survivor Network’,IEEE Transactions on
classifier. These are the dengue attributes of the dataset: Computational Social Systems, 3(2), 75-877)
training phase/testing phase.
SUMMARY 3. Ionuț TARANU (2015),’ Data mining in healthcare:
The prediction of dengue infection carried out using data decision making and precision’, Database Systems
mining techniques such as Association Rule Mining and graph Journal vol. VI, no. 4.
based decision tree. Fever based dataset to be load and finding
the patients affected by dengue fever prediction by machine 4. Khaleel, M. A., Pradham, S. K., & Dash, G.
learning techniques. Association rule mining algorithm useful N(2013),’A survey of data mining techniques on
for analyzing, predicting patient’s behaviors. In the proposed medical data for finding locally frequent
work, a graph based decision tree is proposed to predict the diseases’, International Journal of Advanced Research
dengue fever earlier and reduce mortality rate and classify in Computer Science and Software Engineering, 3(8).
different activities of patients in more accurate manner .This
work proceeds to discuss with predicting patient’s behaviors 5. Kumar, R. N., & Kumar, M. A (2016),’Medical Data
as whether the patients are affected with dengue fever or not. Mining Techniques for Health Care
In addition to this, several prescription drugs used for the Systems’, International Journal of Engineering
management of chronic disease act adversely in dengue Science, 3498.
patients causing further complications. So, the patients
affected with dengue fever are divided into those who are 6. Ling Chen et al (2016)’Personal health indexing
affected with dengue fever, or dengue fever with chronic based on medical examinations’,A data mining
disease warning signs. Overall, the proposed system will approach.
provide a new direction of predicting dengue with chronic
disease through data mining. 7. Mazaher Ghorbani and Masoud Abessi(2017),’ A
New Methodology for Mining Frequent Itemsets on
Temporal Data’, IEEE TRANSACTIONS ON
ENGINEERING MANAGEMENT(2017).
 14 Shikha Bhardwaj et al(2015),’ Improved Apriori
8. Premarathne et al(2016),’Hybrid cryptographic access Algorithm For Association Rules’, International
control for cloud-based EHR systems’. IEEE Cloud Journal of Technical Research and Applications e-
Computing, 3(4), 58-64. ISSN: 2320-8163, www.ijtra.com Volume 3, Issue 3
PP. 238-240.
15.Shikha Bhardwaj et al(2015),’ Improved
9. Purushottam Sharma(2015),’Data Mining Techniques Apriori Algorithm For Association Rules’,
on Medical Data for Finding Locally Frequent International Journal of Technical Research
Diseases’,INTERNATIONAL JOURNAL FOR and Applications e-ISSN: 2320-8163,
RESEARCH IN APPLIED SCIENC E AND www.ijtra.com Volume 3, Issue 3 PP. 238-
ENGINEERING TECHNOLO GY (IJRAS ET) Vol. 240.
2 Issue V, May 2015 Shashi Chhikara1.

10 Rasitha Banu.G and Baviya.M(2015),’Predicting

Thyroid Disease Using Datamining’, International
Journal of Modern Trends i`n Engineering and
Research .

Rituja and Bibave1 , DB. L. Gunjal(2017),’Survey on

11 Mining Health Examination Records’,A Graph-based
Approach, IJARIIE-ISSN(O)-2395-4396 Vol-3 Issue-

12 Seema et al, Decision Tree(2012),’Data Mining

Techniques’,In:International Journal of Latest Trends
in Engineering and Technology (IJLTET).

Simmi Bagga and G.N. Singh(2012),’Applications of

13 Data Mining’, International Journal Science and
emerging Technologies with Latest Trends