BIG DATA ANALYSIS FOR HEART

DISEASE DETECTION SYSTEM

USING MAP REDUCE TECHNIQUE
1 2
G.Vaishali , V.Kalaivani
1
PG Scholar, National Engineering College, Kovilpatti, Tamilnadu,
2
Professor, Department of Computer Science and Engineering, National Engineering College,
Kovilpatti, Tamilnadu
1
vaishvaishu232@gmail.com
2
kalai.nec@gmail.com

Abstract: In today’s world, the enormous information in health care is to be processed in order to identify,
diagnose, detect and prevent the various diseases. Big data analysis is the challenging one because big data
contain large amount of records. It is proposed to develop a centralized patient monitoring system using big
data. In the proposed system, large set of medical records are taken as input. From this medical dataset, it is
aimed to extract the needed information from the record of heart patients using map reduce technique.
Heart disease is a major health problem and it is the leading causes of death throughout the world. Early
detection of heart disease has become an important issue in the medical research fields. For heart disease
detection, some features are analyzed such as RR interval, QRS interval and QT interval. The classification
process states whether the patient is normal or abnormal and in the detection step using map reduce
technique to detect the disease and reduce the dataset. Thus, the proposed system helps to classify a large
and complex medical dataset and detect the heart disease.

Keywords :

Big data, Hadoop and Map Reduce.

I. INTRODUCTION user friendly tool.

Nowadays, a large volume of data is available
in most of the real time applications. This raw
unstructured data is of no use until it is pre-
processed into useful information. It is
necessary to analyze this huge amount of data
and extract useful information from it. For that
extraction process, data mining technology is
needed. Nowadays big data is the emerging
technology. Big data means when the data
which is to be mined varies from a small data
set to a large dataset. It is proposed to develop
the big data analysis for medical application
using data mining. Data mining plays a vital
role in big data.
The term “Big Data” was introduced
in 2005. A massive volume of both structured
and unstructured data and that is so large and it
is difficult to store, analyze, process, share,
visualize and manage with normal database
and software techniques because it is not much
capacity to store large data. Nowadays, many
tools are available for processing big data like
Hadoop, MongoDb, Talend, Tableau, pentaho,
Google charts, SAP inMemory etc. In the
proposed work, use Hadoop tool and it is a
In health care field, there are
different types of data are available like in
signals, images etc. In medical field, big data is
an booming factor because in this lot of
research work are emerging for the
classification of diseases. Big data analytics
enables organizations to analyze a mix of
structured, semi-structured and unstructured
data in search of valuable business
information. Big data analytics is the process
of examining large data sets containing a
variety of data types. In the proposed work,
UCI repository datasets is used for analysis
and some real data datasets are considered
from various hospitals and medical care centre.
1.1 Characteristics of Big Data
There are 4V’s in bigdata. They are
Volume
Volume indicates the amounts of data
generated every second. It mainly referred as
quantity of data. The data available through
social website
and sensor networks going to cross from Peta
bytes to Zeta bytes.
Variety
Variety refers to the different types
of data. In the past, researchers only focused
on structured data and data are arranged in
tabular form like financial data, Structured,
semi-structured, unstructured; Text, image,
audio, video, record.
Velocity
Velocity Refers to the speed of data
processed. This is a concept which indicates
the speed at which the data generated and
become historical. Big data is able to handle all
types of data.
Variability
It describes the amount of variance
used in data that kept within the data bank and
refers how they are spread out or closely
clustered within the data set.
II. HADOOP
Hadoop tool is an Apache open source
framework written in java that allows
distributed processing of large datasets across
grouping of computers using simple programs.
For the proposed work, the hadoop tool is
installed using cygwin terminal for user
friendly environment. Hadoop tool is mainly
used for processing large amount of data. In
hadoop, there is distributed storage system for
storing of the data. Hadoop is designed to scale
up from single server to thousands of systems,
each providing local computation and storage.
Map Reduce
Map Reduce is a programming language
for writing parallel and distributed applications
devised at distributed storage level. The
MapReduce program run on the Hadoop tool
in an Apache open source framework. A Map
Reduce program is composed of a Map
function procedure that performs mapping
process and produce intermediate key value
pair as (k1,va1) -> list(k2,v2) and a Reduce
method as (k2,list(v2)) -> (k3, v3) that
performs that match with intermediate key
value pair and display the count value as
output.
III LITERATURE SURVEY
Xindong Wu et al [10] proposed
HACE theorem that characterized the features
of the big data revolution and proposed a big
data processing model from the data mining
perspective. The data-driven model involves
demand driven aggregation of information,
source, mining and analysis, user interest
modeling and security. One of the main
characteristics of big data application is
autonomous data sources with distributed and
decentralized control. In which the authors
analyzed some issues in the tier model as data
sharing and privacy, domain and application
knowledge. Big data is mainly applicable for
healthcare system. Kiyana Zolfaghar et al [3]
discussed about big data driven solutions to
predict the 30-day risk of readmission for
congestive heart failure (CHF) incident. They
mainly used MHS (Multicare Health System)
data set. First they extracted useful factor from
National Inpatient Dataset(NIS) and augment it
with our patient data set from MHS. Then they
developed scalable data mining models to
predict risk of readmission using the integrated
data set. Also they used random forest
algorithm because it can work with all types of
predictor variables. In the above paper, they
discussed 30-day risk solution. For taking the
dataset as whole like rural area, the data
processing may vary. Muni Kumar N et al[4]
identified the massive shortage of proper
healthcare facilities and addressed how to
provide greater access to primary health care
service in rural areas of India. Big data
processing in real time situation is to turn the
dream of Svasth Bharath (Healthy India) into
reality. They analyzed some key factors to
make the performance of health centre better
and people live healthier.
The proposed concept enables
doctors, patients and staff to have role-based
access to information on electronic health
records. They proposed the following seven
big ideas to fix rural health care in India and
bridge the gap between quality and
affordability in government hospitals. For
processing these large volume of data, they
used hadoop tool. K. Sharmila et al [9]
examined and revealed the benefits of hadoop
in the healthcare sector using data mining. The
apache hadoop has become a worldwide
adoption and it has brought parallel processing
in the hands of average programmer for big
data. They presented an overview of various
data mining techniques and application of
diabetic dataset using this platform. This
helped us to acquire knowledge about how
hadoop can be implemented to predict the
diabetics and related disease. In the hadoop
tool, they have used map reduce concept.
Map reduce can divide the dataset
into multiple chunks, each will be processed in
parallel among multiple nodes. Sathiyavathi R
[8] enlightened that the first step is to collect
the data from various sources, prediction
attribute will be identified and then the
respective algorithm should be applied in the
case. It specifies a map function that processes
a key/value pair to generate a set of
intermediate key/value pairs and a reduce
function that merges all intermediate values
associated with the same intermediate key. For
improving the above model, Saravana N et al
[7] used predictive analysis algorithm in
hadoop/map reduce environment to predict the
diabetes types and the type of treatment to be
provided. This algorithm included various
phases like data warehousing, data collection,
analysis and submitted the analyzed report. For
diabetic treatment, it is necessary to test the
patterns like plasma, glucose concentration,
serum insulin, diastolic blood pressure,
diabetes pedigree, Body Mass Index (BMI).
This system is used to predict and classify the
types of DM and it leads to the improved focus
on every individual patient health. In another
research paper, A.Pradeepa et al [5] presented
consequent algorithms corresponding to the
map reduce based on roughest theory, they put
forward to deal with the massive data. Rough
set theory proposed a new mathematical
approach to imperfect knowledge. It explained
the topological operations, interior and closure
called approximations. Rough set has resolved
complex problems. It is a powerful
mathematical tool to describe the dependencies
among attributes, evaluate the significance of
attributes, and derive decision rules.
This hadoop tool is also used in
enterprise data, Praveen Kumar et al [6]
discussed the challenges of processing these
huge chunk of data and have found that none
of the existing centralized architecture could
efficiently handle this huge volume of data.
Map reduce is a tool implemented for
managing and processing vast amount of
unstructured data in parallel. Programs which
are map reduces are programmed to manage
the vast amounts of data. This enables parallel
processing of the problem and efficient
computation was possible. For big data
security Bhawna Gupta et al [1] discussed
about some techniques how big data is
analyzed by using the technique of hadoop and
why the big data security analytics is important
to mitigate the security threats to secure the
enterprise data more efficiently.
There should be number of
opportunities for big data security analytics to
enter the enterprise security. They would use
the result for securing and implementing
preventive measures from threats to enterprise
data. Some researchers are using network
monitoring tools like Packet pig, Mahout etc.
to enhance the security levels. Then for
improving the efficiency of map reduce
functionality, Devi.L [2] et al suggested the
concept as cache manager. Before executing
the actual computing work, task queries the
cache manager. In a data aware cache, cache
request and cache reply mechanisms are
designed. Implementing cache by extending
hadoop it improves the completion time of
map reduce jobs. It detects the occurrence of
repeated job in the incremental data process.
Also, it stops the repeated work and minimizes
the processing time so that to provide the
optimized usage of Map Reduce nodes. The
data aware cache in map reduce framework
helps to overcome this problem and provide
high efficiency in incremental processing.
IV PROPOSED WORK
The main objective of the proposed
work is to build a Big data Analysis System
that helps to classify a large and complex
medical dataset and detect the disease. In the
proposed system, large set of medical records
are considered. From this medical dataset, it is
aimed to extract the needed information from
the record of heart disease patients. For this
extraction, features in the data set are analyzed.
The analyzed features are classified to detect
the condition of the heart as normal or
abnormal. Also, it is aimed to identify the type
of heart disease and reduce the dataset using
map reduce concept. The goal is to extract the
useful information from large volumes of
dataset collected from various sources. In the
proposed system, it is aimed to take heart
disease dataset to classify and detect the
various types of heart disease.
Figure 4.1 Flow diagram for the proposed
system
4.1 Feature Analysis
From the preprocessed data, some
features for disease detection are analyzed. The
features to be analyzed are RR interval, QRS
interval, QT interval. In the feature analysis
phase, the mean value for all the intervals are
calculated for analyzing the type of heart
diseases
4.2 Reducing Phase The figure 5.2 shows the feature values
In reducing phase, reduce function is such as RR interval, QRS interval, QT
used to merge the values from map function interval for every patients.
into a single result. It reduces a set of
intermediate values which share a key to a
smaller set of values.
4.3 Mapping Phase
In a mapping phase, first mapper
tokenizes the document and emits an
intermediate key-value pair for every record.
After this process each of these elements will
then be sorted by their key.

V IMPLEMENTATION RESULTS
The following Fig 5.1 shows the GUI
of the proposed system of Centralized Patient
Monitoring System using Big Data.

Figure 5.3 Shows the feature value for per

day
Figure 5.3 shows, by clicking per day
button, it shows the mean value for above
selected interval for the patient in the next
frame.

Figure 5.1 Front page of the proposed

system

5.2 Feature Analysis

Fig 5.4 Shows the feature value for per week

Figure 5.4 shows, by clicking per

week button, it shows the mean value for
above selected interval for the patient in the
Fig. 5.2 Displaying the feature values for next frame.
every patient.
5.5 Patient entry form
Figure 5.5 Patient entry form in the system
Figure 5.5 shows, by clicking the
insert button, the patient details are entered and
it is inserted successfully in the database.
5.6 Identification of Disease
Figure 5.6 Checking of patient condition
Figure 5.6 shows, by clicking get
status button in the frame, it displays the
patient condition and it displays the type of
class.
5.7 Classification Results
Figure 5.7 Classification result
Figure 5.7 shows, by clicking the
button ‘Get status’ it displays the patient
condition and class.
Thus the implementation results show the
details of heart patient, it displays the patient
condition and it displays the classification
result of heart disease of the patient.
6. CONCLUSION
Thus the data set are preprocessed and
features are analyzed. The features of heart
disease dataset such as RR interval, QRS
interval and QT interval are analyzed. Using
rule based classification, the features are
classified for knowing the patient’s condition
and it displays the type of heart disease. In the
future work, the data set will be reduced using
map reduce technique. This system is expected
to be useful in the medical field for the
physician to easily analyze the heart disease. It
will aid the physicians for taking decision.
7. FUTURE WORK
For future work, we will detect many
type of diseases using the big data set and by
analyzing the data set, the required data will be
predicted from the data set as easily using map
reduce concept.
8. REFERENCES
[1] Bhawna Gupta and Dr. Kiran Jyoti
(2014),” Big Data Analytics with Hadoop to
analyze Targeted Attacks on Enterprise Data”,
(IJCSIT) International Journal of Computer
Science and Information Technologies, Vol. 5,
No.3, pp. 3867-3870.

[2] Devi. L and S. Gowri (2015),” Optimizing

MapReduce Functionality in bigdata using
Cache Manager”, ARPN Journal of
Engineering and Applied SciencesVol. 10,
No.12, pp. 1819-6608.

[3] Kiyana Zolfaghar, Naren Meadem, Ankur

Teredesai, Senjuti Basu Roy, Si-Chi Chin and
Brian Muckian(2013) “Big Data Solutions for
Predicting Risk-of-Readmission for
Congestive Heart Failure”, IEEE International
Conference on Big Data Vol.3, No 6, pp. 64-
71.

[4] Muni Kumar N and Manjula R (2014),

“Role of Big Data Analytics in Rural Health
Care -A Step Towards Svasth Bharath” ,
(IJCSIT) International Journal of Computer
Science andInformation Technologies, Vol.5,
No.6, pp.7172-7178.

[5] Pradeepa A, Dr. Antony Selvadoss

Thanamani (2013),” Hadoop File System and
Fundamental Concept of MapReduce Interior
and Closure Rough Set Approximations”,
International Journal of Advanced Research in
Computer and Communication Engineering
Vol. 2 ,No.10, pp. 5865-5868.

[6] Praveen Kumar and Dr Vijay Singh

Rathore (2014.),” Efficient Capabilities of
Processing of Big Data using Hadoop Map
Reduce” , International Journal of Advanced
Research in Computer and Communication
Engineering Vol. 3, No. 6, pp. 4421-4425.

[7] Saravana N ,M Ramachandran and S

Lavanya Kumar (2015),” Predictive
Methodology for Diabetic Data Analysis in
Big Data” , ScienceDirect - Procedia Computer
Science Vol 50, pp. 203 – 208.

[8] Sathiyavathi R (2015), “A Survey: Big

Data Analytics on Healthcare System” ,
HIKARI Ltd Contemporary Engineering
Sciences, Vol. 8, No. 3, pp. 121 – 125.

[9] Sharmila R and Dr. S. A.Vethamanickam

(2015), “Survey on Data Mining Algorithm
and Its Application in Healthcare Sector Using
Hadoop Platform” , International Journal of

[10] Xindong Wu, Fellow, Xingquan

Zhu,Gong-Qing Wu, and Wei Ding (2014)
“Data Mining with Big Data”, IEEE
Transactions on Knowledge and Data
Engineering, Vol. 26, No 1, pp.97-107.