International Journal of Engineering & Technology, 7 (2.

8) (2018) 684-687

International Journal of Engineering & Technology

Website: www.sciencepubco.com/index.php/IJET

Research Paper

Heart disease prediction using machine learning techniques

: a survey
V.V. Ramalingam*, Ayantan Dandapath, M Karthik Raja
1,2Department of Computer Science & Engineering, SRM Institute of Science and Technology
*Corresponding Author E-mail: [email protected]

Abstract

Heart related diseases or Cardiovascular Diseases (CVDs) are the main reason for a huge number of death in the world over the last few
decades and has emerged as the most life-threatening disease, not only in India but in the whole world. So, there is a need of reliable,
accurate and feasible system to diagnose such diseases in time for proper treatment. Machine Learning algorithms and techniques have
been applied to various medical datasets to automate the analysis of large and complex data. Many researchers, in recent times, have
been using several machine learning techniques to help the health care industry and the professionals in the diagnosis of heart related
diseases. This paper presents a survey of various models based on such algorithms and techniques andanalyze their performance. Models
based on supervised learning algorithms such as Support Vector Machines (SVM), K-Nearest Neighbour (KNN), NaïveBayes, Decision
Trees (DT), Random Forest (RF) and ensemble models are found very popular among the researchers.

Keywords: Cardiovascular Diseases; Support Vector Machines; K- Nearest Neighbour; Naïve Bayes; Decision Tree; Random Forest; Ensemble Models.

1. Introduction 2. Dimensionality Reduction

Heart is an important organ of the human body. It pumps blood to Dimensionality Reduction involves selecting amathematical
every part of our anatomy. If it fails to function correctly, then the representation such that one can relate the majority of, but not all,
brain and various other organs will stop working, and within few the variance within the given data, thereby including only most
minutes, the person will die. Change in lifestyle, work related significant information. The data considered for a task or a
stress and bad food habits contribute to the increase in rate of problem, may consists of a lot of attributesor dimensions, but not
several heart related diseases. all of these attributes may equally influence the output. A large
Heart diseases have emerged as one of the most prominent cause number of attributes, or features, may affect the computational
of death all around the world. According to World Health complexity and may even lead to overfitting which leads to poor
Organisation, heart related diseases are responsible for the taking results. Thus, Dimensionality Reduction is a very important step
17.7 million lives every year, 31% of all global deaths. In India considered while building any model. Dimensionality Reduction is
too, heart related diseases have become the leading cause of generally achieved by two methods -Feature Extraction and Feature
mortality [1]. Heart diseases have killed 1.7 million Indians in Selection.
2016, according to the 2016 Global Burden of Disease Report,
released on September 15,2017. Heart related diseases increase A. Feature Extraction
the spending on health care and also reduce the productivity of an In this, a new set of features is derived from the original feature
individual. Estimates made by the World Health Organisation set.Feature extraction involves a transformation of the features.
(WHO), suggest that India have lost up to $237 billion, from This transformation is often not reversible asfew, or maybe many,
2005-2015, due to heart related or Cardiovascular diseases [2]. useful information is lost in the process.In [3]and[4]Principal
Thus, feasible and accurate prediction of heart related diseases is Component Analysis (PCA)is used for feature extraction.
very important. Principal Component Analysis is a popularly used linear
Medical organisations, all around the world, collect data on transformation algorithm. In the feature space, it finds the
various health related issues. These data can be exploited using directions that maximize variance and finds directions that are
various machine learning techniques to gain useful insights. But mutually orthogonal. It is a global algorithm that gives the best
the data collected is very massive and, many a times, this data can reconstruction.
be very noisy. These datasets, which are too overwhelming for
human minds to comprehend, can be easily explored using various B. Feature Selection
machine learning techniques. Thus, these algorithms have become In this, a subset of original feature set is selected. In [5],key
very useful, in recent times, to predict the presence or absence of features are selected by CFS(Correlation based Feature Selection)
heart related diseases accurately. Subset Evaluation combined with Best First Search method to
reduce dimensionality. In [6]chi-square statistics test is used to
select the most significant features.

Copyright © 2018 Authors. This is an open access article distributed under the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
International Journal of Engineering & Technology 685

3. Algorithms and Techniques Used C. K – Nearest Neighbour

In 1951, Hodges et al. introduced a nonparametric technique for
A. Naïve Bayes pattern classification which is popularly known the K-Nearest
Naive Bayes is a simple but an effective classification technique Neighbour rule[13]. K-Nearest Neighbour technique is one of the
which is based on the Bayes Theorem. It assumes independence most elementary but very effective classification techniques. It
among predictors, i.e., the attributes or features should be not makes no assumptions about the data and is generally be used for
correlated to one another or should not, in anyway, be related to classification tasks when there is very less or no prior knowledge
each other. Even if there is dependency, still all these features or about the data distribution. This algorithm involves finding the k
attributes independently contribute to the probability and that is nearest data points in the training set to the data point for which a
why it is called Naïve. target value is unavailable and assigning the average value of the
found data points to it.
In [10] KNN gives an accuracy of 83.16% when the value of k is
equal to 9 while using 10-cross validation technique. In [14]
KNN with Ant Colony Optimization performs better than other
techniques with an accuracy of 70.26% and the error rates is
0.526.Ridhi Saini et al. have obtained a efficiency of 87.5% [15],
which is very good.

D. Decision Tree
Decision tree is a of supervised learning algorithm.This technique
is mostly used in classification problems. It performs effortlessly
withcontinuous and categorical attributes. This algorithm
dividesthe population into two or more similar sets based on the
most significantpredictors.Decision Treealgorithm, first calculates
the entropy of each and every attribute. Then the dataset is split
with the help of thevariables or predictors with maximum
information gain or minimum entropy. These two steps are
In [7], Naive Bayes has achieved an accuracy of 84.1584% with performed recursively with the remaining attributes.
the 10 most significant features which are selected using SVM-
RFE (Recursive Feature Elimination) and gain ratio algorithms
whereas in[8],Naive Bayes has achieved an accuracy of 83.49%
when all 13 attributes of the Cleveland dataset[25] are used.

B. Support Vector Machine

Support Vector Machine is an extremely popular supervised
machine learning technique(having a pre-defined target variable)
which can be used as a classifier as well as a predictor. For
classification, it finds a hyper-plane in the feature space that
differentiates between the classes. An SVM model represents the
training data points as points in the feature space, mapped in such
a way that points belonging to separate classes are segregated by a
margin as wide as possible. The test data points are then mapped
into that same space and are classified based on which side of the
margin they fall.

Fig. 2: Decision Tree

In [10]decision tree has the worst performance with an accuracy

of 77.55% but when decision tree is used with boosting technique
it performs better with an accuracy of 82.17%.In [9] decision tree
performs very poorly with a correctly classified instance
Fig. 1: Support Vector Machine percentage of 42.8954% whereas in [16] also uses the same
dataset but used the J48 algorithm for implementing Decision
Shan Xu et al. have used SVM to achieve an accuracy of 98.9% in Trees and the accuracy thus obtained is 67.7% which is less but
People's Hospital dataset [5].In [9], SVM performs the best with still an improvement on the former. Renu Chauhan et al. have
85.7655% of correctly classified instance and in [10] SVM is used obtained an accuracy of 71.43% [17]. M.A. Jabbar et al. have used
with boosting technique to give an accuracy of 84.81%. alternating decision trees with principle component analysis to
HoudaMezrigui et al. have used SVM to attain a f-measure value obtain an accuracy 92.2%[18].Kamran Farooq et al. have achieved
of 93.5617 [11]. In [12] SVM classifies the pixel variation with an the best results on using decision tree-based classifier combined
accuracy of 92.1% helping to identify the affected region with forward selection which achieves a weighted accuracy of
accurately. 78.4604% [19].
686 International Journal of Engineering & Technology

E. Random Forest diseases but still there is a lot scope of research to be done on how
Random Forest is also a popularly supervised machine learning to handle high dimensional data and overfitting. A lot of research
algorithm.This technique can be used for both regression and can also be done on the correct ensemble of algorithms to use for a
classification tasks but generally performs better in classification particular type of data.
tasks. As the name suggests, Random Forest technique considers
multiple decision trees before giving an output. So, it is basically 5. Acknowledgment
an ensemble of decision trees. This technique is based on the
belief that more number of trees would converge to the right
We sincerely thank the staff of SRM Institute of Science and
decision. For classification, it uses a voting system and then
Technology, that have provided their immense support and
decides the class whereas in regression it takes the mean of all the
guidance throughout the project.
outputs of each of the decision trees. It works well with large
datasets with high dimensionality.
Reference
[1] Ramadoss and Shah B et al.“A. Responding to the threat of
chronic diseases in India”. Lancet. 2005; 366:1744–1749. doi:
10.1016/S0140-6736(05)67343-6.
[2] Global Atlas on Cardiovascular Disease Prevention and Control.
Geneva, Switzerland: World Health Organization, 2011
[3] Dhomse Kanchan B and Mahale Kishor M. et al. “Study of
Machine Learning Algorithms for Special Disease Prediction
using Principal of Component Analysis”, 2016 International
Conference on Global Trends in Signal Processing, Information
Computing and Communication.
[4] R.Kavitha and E.Kannan et al. “An Efficient Framework for
Heart Disease Classification using Feature Extraction and Feature
Selection Technique in Data Mining “, 2016
[5] Shan Xu ,Tiangang Zhu, Zhen Zang, Daoxian Wang, Junfeng Hu
and Xiaohui Duan et al. “Cardiovascular Risk Prediction Method
Based on CFS Subset Evaluation and Random Forest
Classification Framework”, 2017 IEEE 2nd International
Conference on Big Data Analysis.
[6] Manpreet Singh, Levi Monteiro Martins, Patrick Joanis and Vijay
Fig. 3: Random Forest K. Mago et al. “ Building a Cardiovascular Disease Predictive
Model using Structural Equation Model & Fuzzy Cognitive Map”,
In [5], random forest performs exceptionally well. In Cleveland 978-1-5090-0626-7/16/$31.00 c 2016 IEEE.
dataset, random forest has a significantly higher accuracy of [7] Kanika Pahwa and Ravinder Kumar et al. “Prediction of Heart
91.6% than all the other methods. In People's Hospital dataset, it Disease Using Hybrid Technique For Selecting Features”, 2017
achieves an accuracy of 97%. In [20] random forest has achieved 4th IEEE Uttar Pradesh Section International Conference on
an f-measure of 0.86. In [21], random forest is used to predict Electrical, Computer and Electronics (UPCON).
[8] Seyedamin Pouriyeh, Sara Vahid, Giovanna Sannino,
coronary heart disease and it obtains an accuracy of 97.7%. Giuseppe De Pietro, Hamid Arabnia, Juan Gutierrez et al. “ A
Comprehensive Investigation and Comparison of Machine
F. Ensemble Model Learning Techniques in the Domain of Heart Disease”, 22nd IEEE
In ensemble modeling two or more related but different analytical Symposium on Computers and Communication (ISCC 2017):
models are used and produce their results are combined into a Workshops - ICTS4eHealth 2017
single score. [9] Hanen Bouali and Jalel Akaichi et al. “Comparative study of
TahiraMahboob et al. [22] have used an ensemble of SVM, KNN Different classification techniques, heart Diseases use Case.”,
2014 13th International Conference on Machine Learning and
and ANN to achieve an accuracy of 94.12%. The Majority vote- Applications
based model as demonstrated by Saba Bashir et al. [23] which [10] Seyedamin Pouriyeh, Sara Vahid, Giovanna Sannino, Giuseppe
comprises of Naïve Bayes, Decision Tree and Support Vector De Pietro, Hamid Arabnia, Juan Gutierrez et al. “ A
Machine classifiers, gave an accuracy of 82%, sensitivity of 74% Comprehensive Investigation and Comparison of Machine
and specificity of 93% for UCI heart disease dataset. In [24] an Learning Techniques in the Domain of Heart Disease”, 22nd IEEE
ensemble model, consisting ofGini Index,SVMand Naïve Bayes Symposium on Computers and Communication (ISCC 2017):
classifiers, has been proposed which gave an accuracy of 98% in Workshops - ICTS4eHealth 2017
predicting Syncope disease. [11] Houda Mezrigui, Foued Theljani and Kaouther Laabidi et al.
“Decision Support System for Medical Diagnosis Using a Kernel-
Based Approach”, ICCAD’17, Hammamet - Tunisia, January 19-
4. Conclusion 21, 2017.
[12] Dr.(Mrs).D.Pugazhenthi, Quaid-E-Millath and Meenakshi et al.
“Detection Of Ischemic Heart Diseases From Medical Images “
Based on the above review, it can be concluded that there is a 2016 International Conference on Micro-Electronics and
huge scope for machine learning algorithms in predicting Telecommunication Engineering.
cardiovascular diseases or heart related diseases. Each of the [13] J. Hodges et al. “Discriminatory analysis, nonparametric
above-mentioned algorithms have performed extremely well in discrimination: Consistency properties,” 1981.
some cases but poorly in some other cases. Alternating decision [14] S.Rajathi and Dr.G.Radhamani et al. “Prediction and Analysis of
trees when used with PCA, have performed extremely well but Rheumatic Heart Disease using kNN Classification with ACO “,
decision trees have performed very poorly in some other cases 2016.
[15] Puneet Bansal and Ridhi Saini et al. “Classification of heart
which could be due to overfitting. Random Forest and Ensemble
diseases from ECG signals using wavelet transform and kNN
models have performed very well because they solve the problem classifier”, International Conference on Computing,
of overfitting by employing multiple algorithms (multiple Communication and Automation (ICCCA2015).
Decision Trees in case of Random Forest). Models based on Naïve [16] Simge EKIZ and Pakize Erdogmus et al. “Comparitive Study of
Bayes classifier were computationally very fast and have also heart Disease Classification”, 978-1-5386-0440-3/17/$31.00
performed well.SVM performed extremely well for most of the ©2017 IEEE.
cases. Systems based on machine learning algorithms and [17] Renu Chauhan, Pinki Bajaj, Kavita Choudhary and Yogita Gigras
techniques have been very accurate in predicting the heart related et al. “Framework to Predict Health Diseases Using Attribute
International Journal of Engineering & Technology 687

Selection Mechanism”, 2015 2nd International Conference on

Computing for Sustainable Global Development (INDIA Com).
[18] M.A.JABBAR , B.L Deekshatulu and Priti Chndra et al.
“Alternating decision trees for early diagnosis of heart disease”,
Proceedings of International Conference on Circuits,
Communication, Control and Computing (I4C 2014).
[19] Amir Hussain, Peipei Yang, Mufti Mahmud and Jan Karasek et al.
“A Novel Cardiovascular Decision Support Framework for
effective clinical Risk Assessment.”, 978-1-4799-4527-
6/14/$31.00 ©2014 IEEE.
[20] Quazi Abidur Rahman, Larisa G. Tereshchenko, Matthew
Kongkatong, Theodore Abraham, M. Roselle Abraham,
and Hagit Shatkay et al. “Utilizing ECG-based Heartbeat
Classification for Hypertrophic Cardiomyopathy Identification”,
DOI 10.1109/TNB.2015.2426213, IEEE Transactions on Nano
Bioscience TNB-00035-2015.
[21] Ahmad Shahin, Walid Moudani, Fadi Chakik, Mohamad Khalil et
al. ”Data Mining in Healthcare Information Systems: Case Studies
in Northern Lebanon”, ISBN: 978-1-4799-3166-8 ©2014 IEEE.
[22] Tahira Mahboob, Rida Irfan and Bazelah Ghaffar et al.
“Evaluating Ensemble Prediction of Coronary Heart Disease using
Receiver Operating Characteristics”, 978-1-5090-4815-
1/17/$31.00 ©2017 IEEE.
[23] Saba Bashir, Usman Qamar, M.Younus Javed et al. “An Ensemble
based Decision Support Framework for Intelligent Heart Disease
Diagnosis” International Conference on Information Society (i-
Society 2014).
[24] Ammar Asjad Raja, Irfan-ul-Haq , Madiha Guftar Tamim Ahmed
Khan and Dominik Greibl et al. “Intelligent Syncope Disease
Prediction Framework using DM-Ensemble Techniques”, FTC
2016 - Future Technologies Conference 2016.
[25] CI Education, Heart Disease Data Set [OL].
http://archive.ics.uci.edu/ml/datasets/Heart+Disease CHDD.
[26] T. Padmapriya and V.Saminadan, “Handoff Decision for Multi-
user Multiclass Traffic in MIMO-LTE-A Networks”, 2nd
International Conference on Intelligent Computing,
Communication & Convergence (ICCC-2016) – Elsevier -
PROCEDIA OF COMPUTER SCIENCE, vol. 92, pp: 410-417,
August 2016.
[27] S.V.Manikanthan and D.Sugandhi “Interference Alignment
Techniques For Mimo Multicell Based On Relay Interference
Broadcast Channel” International Journal of Emerging
Technology in Computer Science & Electronics (IJETCSE) ISSN:
0976-1353 Volume- 7, Issue 1 –MARCH 2014.