PREDICTIVE ANALYTICS IN HEALTHCARE SYSTEM USING DATA MINING TECHNIQUES

David C. Wyld et al. (Eds) : DBDM, CICS, CSIP, AI&FL, SCOM, CSE, CCNET-2016
pp. 01–09, 2016. © CS & IT-CSCP 2016 DOI : 10.5121/csit.2016.60501
PREDICTIVE ANALYTICS IN
HEALTHCARE SYSTEM USING DATA
MINING TECHNIQUES
Basma Boukenze1*
, Hajar Mousannif2
and Abdelkrim Haqiq3
1
Computer, Networks, Mobility and Modeling laboratory
FST, Hassan 1st University, Settat, Morocco
basma.boukenze@gmail.com
2
LISI Laboratory,FSSM Cadi Ayyad University,Marrakech 40000,Morocco
mousannif@uca.ac.ma
3
Computer, Networks, Mobility and Modeling laboratory
FST, Hassan 1st University, Settat, Morocco
e-NGN Research Group, Africa and Middle East
ahaqiq@gmail.com
ABSTRACT
The health sector has witnessed a great evolution following the development of new computer
technologies, and that pushed this area to produce more medical data, which gave birth to
multiple fields of research. Many efforts are done to cope with the explosion of medical data on
one hand, and to obtain useful knowledge from it on the other hand. This prompted researchers
to apply all the technical innovations like big data analytics, predictive analytics, machine
learning and learning algorithms in order to extract useful knowledge and help in making
decisions. With the promises of predictive analytics in big data, and the use of machine learning
algorithms, predicting future is no longer a difficult task, especially for medicine because
predicting diseases and anticipating the cure became possible. In this paper we will present an
overview on the evolution of big data in healthcare system, and we will apply a learning
algorithm on a set of medical data. The objective is to predict chronic kidney diseases by using
Decision Tree (C4.5) algorithm.
KEYWORDS
big data, big data analytics, machine learning, Healthcare, learning algorithm, C4.5
1. INTRODUCTION
There is an almost universal definition shared with proponents of the ideology of big data, it is
that "Big Data sets a situation in which data sets have increased at such huge sizes that
conventional technologies of information, can no longer manage them effectively, either the size
or the extent and the growth of the data set" [1].
The world has become submerged by a large amount of data. Every moment is equivalent to the
generation of thousands data. All sectors and all their activities are involved due to digitization,

2 Computer Science & Information Technology (CS & IT)
the introduction of information technology as an effective tool, and the Internet which is
becoming a very important user interface for daily interactions [2]. However, these generated
data become more and more difficult to manage in terms of volume, variety and velocity [3]. This
gave birth to a new domain named big data. In 2008, Gartner used for the first time the term "Big
Data", in reference to the explosion of digital data and quoted it will impact the way we work [4].
"Big Data" and "analysis of big data" are inseparable. This reflects the common opinion that "Big
data" does not refer to the problem of information overload, but refers also to the analytical tools
used to manage the flow of data and transform the flood in a source of useful information.
The medical field has its great contribution in this deluge of data because of some technological
innovations in the field like cloud computing which has relocated the tests of care beyond the
four walls of the hospital, and has made them available anywhere and anytime [5], laparoscopic
surgery and robotic surgery, which replaced classical surgery [6], also smart homes which allow
patients self-care and monitoring using simple devices that deliver results on specific
physiological conditions. There are also smart applications or software that can analyze the body
signals using integrated sensors with the aim of monitoring [7], as well as mHealth technologies
that support new methods of biological, behavioural and environmental data collection. These
include sensors that monitor the phenomena with a higher accuracy [8].
All these innovations participated to the explosion of medical data, by multiplying data sources
and electronic medical records containing diagnostic Images, lab results, and biometric
information that are generated and stored [8.9.10].
Researchers have deduced that this explosion of medical data has the potential to improve clinical
decisions at the point of care. Doctor will become able to extract relevant knowledge for each
patient, which gives better decisions and results [11].
On most of this, the term "analyzing medical data" and "predictive analytics" in Google Trends
showed an impressive growth of interest from 2011 [12], because the process of analysis in the
medical sector does not stop just at the level of the ability to manage large databases, but it
exceeds this to the ability to retrieve future knowledge, which is encouraged by many researchers
and experts. Seen that an analysis of the big data shows itself as the only solution able to solve all
the problems of the medical sector [13] by:
Providing best Service
Monitoring quality in hospital
Improving treatment processes
Detection of disease earlier
There are many algorithms for classification and prediction applied to predict the most killer
diseases like breast cancer, heart disease, motor neuron, and diabetes. In this present paper, we
apply a decision tree classifier (C4.5) [14], which is among the most influential data mining
algorithm in the research community and among the top 10 data mining algorithms. Our aim is to
predict chronic kidney disease by this learning algorithm.
The rest of this paper is organized as follows:

Computer Science & Information Technology (CS & IT) 3
• Section 2 is about related work.
• Section 3 presents the context of the experiment, metrics and research hypothesis.
• Section 4 presents Experimental results
• Section 5 discusses the results
Finally, section 6 concludes the paper.
2. RELATED WORKS
Andrew Kusiak et al [15] have used data pre-processing, data transformations, and a data
mining approach to elicit knowledge about the interaction between many of measured parameters
and patient survival. Two different data mining algorithms were engaged for extracting
knowledge in the form of decision rules. Those rules were used by a decision-making algorithm,
which predicts survival of new unseen patients. Important parameters identified by data mining
were interpreted for their medical significance. They have introduced a concept in their research
work have been applied and tested using collected data at four dialysis sites. The approach
presented in their paper reduces the cost and effort of selecting patients for clinical studies.
Patients can be chosen based on the prediction results and the most important parameters
discovered.
Abhishek et.al [16] have used two neural network techniques, Back Propagation Algorithm
(BPA), Radial Basis Function (RBF) and one non-linear classifier Support Vector Machine
(SVM) and compared in accordance with their efficiency and accuracy. They used WEKA 3.6.5
tool for implementation to find the best technique among the above three algorithms for Kidney
Stone Diagnosis. The main purpose of their thesis work was to propose the best tool for medical
diagnosis, like kidney stone identification, to reduce the diagnosis time and improve the
efficiency and accuracy. From the experimental results they concluded, the back propagation
(BPA) significantly improved the conventional classification technique for use in medical field.
Andrew Kusiak et al [17] have used data pre-processing, data transformations, and data mining
approach to elicit knowledge about the interaction between many of measured parameters and
patient survival. Two different data mining algorithms were employed for extracting knowledge
in the form of decision rules. Those rules were used by a decision-making algorithm, which
predicts survival of new unseen patients. Important parameters identified by data mining were
interpreted for their medical significance. They have introduced a new concept in their research
work, it has been applied and tested using collected data at four dialysis sites. The approach
presented in their paper reduced the cost and effort of selecting patients for clinical studies.
Patients can be chosen based on the prediction results and the most significant parameters
discovered.
Ashfaq Ahmed K et.al, [18] have presented a work using machine learning techniques, namely
Support Vector Machine [SVM] and Random Forest [RF]. These were used to study, classify and
compare cancer, liver and heart disease data sets with varying kernels and kernel parameters.
Results of Random Forest and Support Vector Machines were compared for different data sets
such as breast cancer disease dataset, liver disease dataset and heart disease dataset. The results
with different kernels were tuned with proper parameter selection. Results were better analyzed to
establish better learning techniques for predictions. It is concluded that varying results were
observed with SVM classification technique with different kernel functions.

Sadik Kara et.al [19] had concentrated on the diagnosis of optic nerve disease through the
analysis of pattern electroretinography (PERG) signals with the help of artificial neural network
(ANN). Implemented Multilayer feed forward ANN trained with a Levenberg Marquart
(LM) back propagation algorithm. The end results were classified as healthy and diseased. The
stated results shown that the proposed method PERG could make an effective interpretation.
With respect to all related work mentioned above, our work is predicting disease using chronic
kidney failure datasets by C4.5 algorithm.
3. EXPERIMENT
In this work, we will apply C4.5, a learning algorithm that will make classification and prediction
on a database to extract knowledge and classify patients into two categories: chronic kidney
disease (ckd) and not chronic kidney disease (notckd).
3.1 Experiment environment:
In this study, we use the Waikato Environment for Knowledge Analysis (Weka). It is a
comprehensive suite of Java class libraries that implement many algorithms for data mining
clustering, classification, regression, analysis of results. This platform gives to researchers a
perfect environment to implement and evaluate their classification model comparing to
TANAGRA or ORANGE [20]
3.2 Chronic kidney disease dataset
We used the database Chronic Kidney Disease Dataset from UCI Machine Learning Repository
[21]. This database contains 400 instances and 24 integer attributes, two class (chronic kidney
disease (ckd), not chronic kidney disease (notckd)). Table 1 describes the attributes of the
database, while Table 2 describes the distribution of classes.
Table1 : Information attribute
Attribute Representation Information
attribute
Description
Age Age Numerical Years
Blood pressure Bp Numerical Mm/Hg
Specific gravity Sg Nominal 1.005,1.010,1.015,1.020,1.025
Albumin Al Nominal 0.1.2.3.4.5
Sugar Su Nominal 0.1.2.3.4.5
Red blood cells Rbc Nominal Normal, abnormal
Pus cell Pc Nominal Normal, abnormal
Pus cell clumps Pcc Nominal Present, notpresent
Bacteria Ba Nominal Present, notpresent
Blood glucose
random
Bgr Numerical Mgs/dl
Blood urea Bu Numerical Mgs/dl
Serum creatinin Sc Numerical Mgs/dl
Sodium Sod Numerical mEq/L
Potassium Pot Numerical mEq/L

Haemoglobin Hemo Numerical Gms
Packed cell volume Pcv Numerical
White blood cell
count
Wc Numerical Cells/cumm
Red blood cell
count
Rc Numerical Millions/cmm
Hypertension Htn Nominal Yes, no
Diabetes mellitus Dm Nominal Yes, no
Coronary artery
disease
Cad Nominal Yes, no
Appetite Appet Nominal Good, poor
Pedal edema Pe Nominal Yes, no
Anemia Ane Nominal Yes, no
Class Classe Nominal Ckd notckd
Table 2 : Class distribution
Class Distribution
1 Ckd 250 (62.5%)
2 Notckd 150 (37.5%)
3.3 Metrics and research hypotheses
To understand classifier’s behaviour we should calculate those metrics, we use hypothesis below:
• True positive (TP) is the number of positive samples correctly predicted.
• True negative (TN) is the number of negative samples correctly predicted
• False negative (FN) is the number of positive samples wrongly predicted.
• False positive (FP) is the number of negative samples wrongly predicted as positive.
Table 3. Metric and research hypotheses

Another important metric Confusion Matrix is taken into account. It is a visualization tool which
is commonly used to present the accuracy of the classifiers in classification. The columns
represent the predictions, and the rows represent the actual class as shown in Table 4
Table 4. Confusion matrix description.
4. EXPERIMENTAL RESULTS
In order to apply our classifier and evaluate its performance, we apply the 10-fold cross
validation test which is a technique that splits the original set into a training sample to train the
model, and a test set to evaluate it. After applying the pre-processing and preparation methods,
we try to analyse the data visually and figure out the distribution of values in terms of
performance and accuracy of the model.
Table 5 : C4.5 performance
Table 6 : Simulation error
Evaluation
criteria
C4.5
Time to build
model (s)
0.08
Correctly
classified
instances
396
Incorrectly
classified
instance
4
Accuracy 63%
Error 0.37
Evaluation criteria C4.5
Kappa statistic 0.97
Mean absolute error 0.02
Root mean squared error 0.08
Relative absolute error % 4.79
Root relative squared error % 16.66

Table 7 : Accuracy measures by class
Table 8 : Diffusion Matrix
Ckd NotCkd
C4.5 (J48)
249
3
1
147
Ckd
Notckd
5. DISCUSSION
we can say that the C4.5 classifier is powerful, according to the number of correctly classified
instances (396) and just 4 instances misclassified, This is mentioned with a low error rate (0.37 )
(see Table 5) . We can also deduce that this algorithm is excellent because of the value of KS =
0.97, which reflects the performance and accuracy of the classifier (see table 6). Fromtable 7, we
deduce that C4.5 achieves best results regarding precision (0.98 ckd and 0.99 notckd), and
sensitivity or (recall) (0.99ckd and 0.98 notckd ).
C4.5 has proved its performance as a powerful classifier in term of accuracy and the minimum
execution time, which makes it a good classifier to be used in the medical field for classification
and prediction.
6. CONCLUSION
As conclusion, the application of data mining techniques for predictive analysis is very important
in the health field because it gives us the power to face diseases earlier that threaten the human
being; child, young and old people, through the anticipation of cure and helping in decision-
making. In this work we used a learning algorithm C4.5 to predict patients with chronic kidney
failure disease (ckd), and patients who are not suffering from this disease (notckd). The classifier
used proved its performance in predicting with best results in terms of accuracy and minimum
execution time.
This is a major challenge in the medical field and pushes us to increase our efforts to develop
machine learning methods, to exploit information intelligently and extract the best knowledge.
REFERENCES
[1] franck Ohlhorst, January 2013 ’ Big Data Analytics: Turning Big Data into Big Money’, ISBN: 978-
1-118-14759-7, pp 176 .
[2] Samson Oluwaseun, F., Serdar , S., and Vanduhe,V ., (2014)," Advancing big data for humanitarian
needs ", Procedia Engineering, vol . 78,N., pp 88-95
TP FP precision recall F-measure Class
C4.5 0.99
0.98
0.02
0.004
0.98
0.99
0.99
0.98
0.99
0.98
Ckd
Notckd

[3] Amir, G., Murtaza, H., (2015),"Beyond the hype: Big data concepts, methods, and analytics»,
International journal of Information Management, vol . ,pp 137-144.
[4] H., Chen, H. L., Chiang, C., Storey, (2012), ‘BUSINESS INTELLIGENCE AND ANALYTICS:
FROM BIG DATA TO BIG IMPACT’, MIS Quarterly, Vol. 36 ,No. 4, pp. 1165-1188.
[5] Jonathan Northover, Brian McVeigh, Sharat Krishnagiri. Healthcare in the cloud: the opportunity
and the challenge. MLD.
Available at http://www.sunquestinfo.com/images/uploads/CMS/445/mlo_02-12014_
healthcare_in_the_cloud.pdf
[6] Gabriel I. Barbas, Sherry A. Glied, (2010),’’ New Technology and Health Care Costs — The Case of
Robot-Assisted Surgery’’; the new England journal of medicine , N°, 363, pp 707-704 . Available at
http://www.nejm.org/doi/full/10.1056/NEJMp1006602
[7] Marianthi Theoharidou, Nikos Tsalis, ‘’Smart Home Solutions for Healthcare: Privacy in Ubiquitous
Computing Infrastructures’’.
Available online at http://www.cis.aueb.gr/Publications/Smart%20Home%20-%20Site%20TR.pdf
[8] Steve G. Peters, James D. Buntrock,(2014),’’Big Data and the Electronic Health Record’’,
Ambulatory Care Manage , Vol. 37, No. 3, pp. 206–210
[9] R. Weil, (2014),’’ Big Data In Health: A New Era For Research And Patient Care Alan R. Weil’’,
Health Affair, Vol. 33, N° 7, pp 1110.
[10] Peter Groves; Basel Kayyali, ( 2013),’’ The ‘big data’ revolution in healthcare’’, McKinsy and
Company. Center for US Health System Reform Business Technology Office. Available at http
://digitalstrategy.nl/wp-content/uploads/E2-2013.04-The-big-data-revolution-in-US-health-care-
Accelerating-value-and-innovation.pdf.
[11] T., Huang, L., Lan, (2015), ‘’Promises and Challenges of Big Data Computing in Health Sciences’’,
Big Data Research vol. 2, pp 2-11 available at http://dx.doi.org/10.1016/j.bdr.2015.02.002
[12] Khurshid R., G., Kai, Z., John T., W., and Charles P., F., (2014), ‘’Harnessing Big Data for Health
Care and Research Are Urologists Ready? ", Journal of European Urology, vol. N., pp 1-3
[13] Wullianallur Raghupathi,Viju Raghupathi, (2014),’’Big data analytics in healthcare: promise and
Potential’’, Health Information Science and Systems.
Available at http://www.biomedcentral.com/content/pdf/2047-2501-2-3.pdf
[14] Rashedur M. Rahman, Fazle Rabbi Md. Hasan ‘Using and comparing different decision tree
classiﬁcation techniques for mining ICDDR, B Hospital Surveillance data’’, Elsevier, Vol. 38, pp
11421–11436
[15] AndrewKusiak, Bradley Dixonb, Shital Shaha, (2005),’’ Predicting survival time for kidney dialysis
patients: a data mining approach’’, Elsevier Publication, Computers in Biology and Medicine, Vol.
35, pp 311–327
[16] Abhishek, Gour Sundar Mitra Thakur, Dolly Gupta, (2012) “Proposing Efficient Neural Network
Training Model for Kidney Stone Diagnosis”, International Journal of Computer Science and
Information Technologies, Vol. 3 (3), pp 3900-3904

[17] AndrewKusiak, Bradley Dixonb, Shital Shaha, (2005) “Predicting survival time for kidney dialysis
patients: a data mining approach”, Elsevier Publication, Computers in Biology and Medicine ,Vol.35,
pp 311–327
[18] Ashfaq Ahmed K, Sultan Aljahdali and Syed Naimatullah Hussain, (2013) “Comparative Prediction
Performance with Support Vector Machine and Random Forest Classification Techniques”,
International Journal of Computer Applications Vol. 69, No.11, pp 12-16
[19] Sadik Kara, Aysegul Guvenb, Ayse OztUrk Onerc, (2006) “Utilization of artificial neural networks in
the diagnosis of optic nerve diseases”, Elsevier Publication, Computers in Biology and Medicine,Vol.
36, pp 428–437
[20] M Hall, E Frank, G Holmes, B Pfahringer,( 2009), ‘The WEKA data mining software: an update’,
Volume 11, Issue 1, pp 10-18
[21] “UCI Machine Learning Repository: Kidney failure Data Set [Online].
Available: https://archive.ics.uci.edu/ml/datasets/Chronic_Kidney_Disease#
AUTHOR
Basma BOUKENZE, PhD student at the Faculty of Science and Technology in
Settat, Morroco. after obtaining a degree in computer Genie in 2009, and a master
degree in engineering network and System in 2011, has continued studies and
currently registered in doctoral science and technology formation at the Faculty of
Science and technology Settat, Moroco, member of The Mathematical Research
structure and applied computing. Laboratory Computer Networks, Mobility and
modelling.

PREDICTIVE ANALYTICS IN HEALTHCARE SYSTEM USING DATA MINING TECHNIQUES

More Related Content

What's hot (20)

Similar to PREDICTIVE ANALYTICS IN HEALTHCARE SYSTEM USING DATA MINING TECHNIQUES (20)

More from cscpconf (20)

Recently uploaded (20)

PREDICTIVE ANALYTICS IN HEALTHCARE SYSTEM USING DATA MINING TECHNIQUES