Biosc.Biotech.Res.Comm.

Special Issue Vol 13 No 11 (2020) Pp-24-27

Heart Disease Prediction Using Machine Learning

Algorithms
Malavika G1, Rajathi N2, Vanitha V3 and Parameswari P4
1
PG Scholar, Department of Information Technology, Kumaraguru College of Technology, Coimbatore, India.
2,3
Professor, Department of Information Technology, Kumaraguru College of Technology, Coimbatore, India.
4
Assistant Professor (SRG) Department of MCA, Kumaraguru College of Technology, Coimbatore, India.

ABSTRACT
The rapidly growing field of data analysis plays a significant role in healthcare. The healthcare industry has become
big business. The healthcare sector produces enormous amounts of data every day. This data helps to extract the hidden
information, which is useful to predict disease at the earlier. In medical field, predicting heart disease is treated as one
of the intricate tasks. Therefore, there is a necessity to develop a decision support system to forecast the cardio vascular
disease in a patient. Machine learning plays a vital part in disease prediction. In this paper, various machine learning
methods were used to predict the heart disease and their performances were compared. The results obtained show the
superiority of the Random forest algorithm.

KEY WORDS: CLASSIFICATION ACCURACY, HEART DISEASE, MACHINE LEARNING.

INTRODUCTION Cardiovascular diseases (CVDs) are the most common

explanation for global death. It is estimated that 17.9
Data mining is used to examines and unearths important million die annually. Heart attacks- once characterized
information from a massive collection of data. This can be as a part of “the old man’s disease” but in this era it can
further helpful in exploratory and illustration out patterns causes for more young people. The heart attack occurs
for making intelligent business-related decisions. One of when the coronary arteries become blocked. It causes
the most threatening in medical domain is heart disease, a serious attack when one or more coronary arteries
which occurs instantly when its limitations are reached. become blocked.
Machine learning plays a vital role in disease prediction
Rajathi N et al,. Cardiovascular disease generally refers Bad clinical results would be the doorway in the death
to narrowed or blocked blood vessels, which can also of a patient. A computer-based support system can be
lead to heart attack, chest pain or stroke. In general, developed to make a good decision in order to achieve
blood pressure, cholesterol and pulse rate are the main correct and cost-effective treatment. Most of the hospitals
reasons for a heart attack. Heart attack is the main heart maintains their patient data in the form of images, texts
disease. and numbers using database systems. This data contains
much of the hidden information that has not yet explored,
which are useful to make right decisions. Therefore,
ARTICLE INFORMATION there is a need to develop an excellent system to help
*Corresponding Author: rajathi.in.it@kct.ac.in the practitioners to predict heart disease before it occurs.
Received 9th Oct 2020 Accepted after revision 7th Dec 2020 This paper mainly concentrates on the prediction of
Print ISSN: 0974-6455 Online ISSN: 2321-4007 CODEN: BBRCBA heart disease considering the past heart disease database
Thomson Reuters ISI Web of Science Clarivate Analytics USA and
records.
Crossref Indexed Journal

24
NAAS Journal Score 2020 (4.31) SJIF: 2020 (7.728)
A Society of Science and Nature Publication,
Bhopal India 2020. All rights reserved.
Online Contents Available at: http//www.bbrc.in/
Doi: http://dx.doi.org/10.21786/bbrc/13.11/6
Malavika G et al.,

MATERIAL AND METHODS Proposed Methodology: In this paper, various machine

learning methods including Naive Bayes classifier,
Various studies with respect to diagnosis of heart disease logistic regression, random forest, support vector
are discussed in this section. Feixiang Huang et. al., machine, decision tree classifier and KNN are employed
used a data mining process to foresee hypertension to forecast heart disease. The Python language is used
from patient medical histories and concluded that for implementation. The working of the model proposed
J-48 classifier produces better results. M. Amiri et. al., is pictorially depicted in Figure 1. The dataset is pre-
developed diagnosis systems heart sounds. They used processed in order to remove irrelevant data which helps
116 heart sound signals to classification and regression to achieve better accuracy.
trees. M.A. Nishara Banu et. al., used clustering and
classification algorithm to forecast the hazard level of Dataset: The heart disease dataset available in UCI
the patients. The authors Theresa Princy et. al.., discussed repository taken for this study. The dataset consisting
about classification methods including Naïve Bayes, of the parameters including age, sex, chest pain type,
neural network, KNN, decision tree for predicting the risk serum cholesterol, resting blood pressure etc. After pre-
level of a patient they consider age, gender, pulse rate, processing the dataset was separated into training (70%)
blood pressure, cholesterol of each patient. and testing (30%). The models used logistic regression, K-
Nearest Neighbor, Support Vector Machine, Naïve Bayes,
The various the machine learning algorithms are used Decision tree and Random forest are trained using the
by Min Chen et. al., for effective prediction of chronic training data and finally tested with the testing set.
disease. A multimodal disease risk prediction method
was adopted for structured and unstructured data. The RESULTS AND DISCUSSION
prediction accuracy the algorithm is better than other
with a convergence speed. Tikotikar A et.al., data mining The overall objective of this paper is to forecast
technique are used in the medical field for clinical more accurately the occurrence of heart disease.
diagnosis. It is inferred that an exhaustive survey of Simulation based experiments were conducted using six
medical data help to make well informed diagnosis and methodologies named Naive Bayes Classifier, Logistic
decisions. Regression, Random Forest, SVM, Decision Tree Classifier
and KNN. From the result it’s been seen that the random
Cincy Raju et. al.., proved that the SVM technique is an forest gives more accuracy as compared as other five
efficient method for predicting heart disease. Praveen techniques. The data set used is decomposed into a
Kumar Reddy. M, et. al., used decision tree algorithm to training set and testing set. Here, 70% of the dataset is
prove the better prediction by comparing its performance taken for training and the remaining is considered for
with SVM. The authors Akash et. al., applied structured testing. From the dataset, it is identified that there are
data and the text data of the patient to the k-mean more people suffering from heart disease in the 50-60
algorithm and archived better accuracy. Reddy et. al., age group. This is pictorially represented in figure 2.
employed machine learning methods for heart disease
prediction. All these created an interest to employ
machine learning to prediction of heart disease. Figure 2: Number of heart patients in different age
group

Figure 1: Proposed System

From the dataset, it is inferred clearly that a greater

number of men are suffering from heart diseases as
compared to women. While the range of men suffering
from heart disease lies between 80-100, the number of
women suffering from heart disease lies between 60-
80. This is shown in figure 3. The performance of the
classification models on the test data was represented
using confusion matrix, per class accuracy and
classification accuracy and is given in table 1.

25
 Malavika G et al.,

Figure 3: Presence of heart disease based on Gender Table 2. Classification Accuracy of Classifiers

Methodology Prediction Accuracy

Logistic Regression 86.88%

K- Nearest Neighbor 86.88%
Support Vector Machine 88.52%
Decision Tree 78.68%
Naïve Bayes 88.52%
Random Forest 91.80%

Figure 4: Performance of Classifiers

Table 1. Classification Performance of Various

Algorithms

Methodology Confusion Matrix Per Class

0 (Female) 1(Male) Accuracy

Logistic Regression 23 4 85.18%

4 30 88.23%
K- Nearest Neighbor 23 4 85.18%
4 30 88.23%
Support Vector 23 4 85.18%
Machine 3 31 91.17%
Decision Tree 21 6 77.77%
7 27 79.41%
Naïve Bayes 24 3 88.88% regression trees. In The 2013 International Joint
4 30 88.23% Conference on Neural Networks (IJCNN) (pp. 1-4).
Random Forest 25 2 92.59% Banu, M.N. and Gomathy, B., 2014, March. Disease
3 31 91.17% forecasting system using data mining methods. In
2014 International conference on intelligent computing
The classification accuracy of various algorithms is applications (pp. 130-133). IEEE.
graphically represented in figure 4 and the results Chen, M., Hao, Y., Hwang, K., Wang, L. and Wang, L.,
are presented in table 2. From the results achieved it
2017. Disease prediction by machine learning over big
is inferred that random forest algorithm gives best
prediction accuracy than other algorithms. data from healthcare communities. IEEE Access, 5,
pp.8869-8879.
CONCLUSION Huang, F., Wang, S. and Chan, C.C., 2012, August.
Predicting disease by using data mining based
In the field of disease prediction, machine learning on healthcare information system. In 2012 IEEE
plays a significant role. In this paper, various machine
International Conference on granular computing (pp.
learning approaches are used for heart disease forecast.
The experimental results illustrate that the Random 191-194).
Forest algorithm achieves the highest accuracy of Jamgade, A.C. and Zade, S.D., 2019. Disease prediction
91.8% and thus successfully achieving the objective of using machine learning. International Research Journal
improving the prediction accuracy. The future work is of Engineering and Technology, 6(5), pp.6937-6938.
towards more investigation on evolutionary computation
Prasad, R., Anjali, P., Adil, S. and Deepa, N., 2019. Heart
techniques for the problem undertaken and study their
performances. disease prediction using logistic regression algorithm
using machine learning. International journal of
REFERENCES Engineering and Advanced Technology, 8, pp.659-
Amiri, A.M. and Armano, G., 2013, August. Early 662.
diagnosis of heart disease using classification and Praveen Kumar Reddy, M., Sunil Kumar Reddy, T.,
26
Malavika G et al.,

Balakrishnan, S., Syed Muzamil Basha, & Ravi Kumar using Data Mining Techniques. In 2018 Conference on
Poluru., 2019. Heart Disease Prediction Using Machine Emerging Devices and Smart Systems (ICEDSS) (pp.
Learning Algorithm. International Journal of Innovative 253-255). IEEE.
Technology and Exploring Engineering (IJITEE) ISSN: Thomas, J. and Princy, R.T., 2016, March. Human heart
2278-3075, Volume-8 Issue-10. disease prediction system using data mining techniques.
Rajathi, N., Kanagaraj, S., Brahmanambika, R. and In 2016 International Conference on Circuit, Power and
Manjubarkavi, K., 2018. Early detection of dengue using Computing Technologies (ICCPCT) (pp. 1-5). IEEE.
machine learning algorithms. International Journal Tikotikar, A., & Kodabagi, M., 2017. A survey on
of Pure and Applied Mathematics, 118(18), pp.3881- technique for prediction of disease in medical data. In
3887. 2017 International Conference on Smart Technologies
Raju, C., Philipsy, E., Chacko, S., Suresh, L.P. and Rajan, for Smart Nation (Smart Tech Con) (pp. 550-555).
S.D., 2018, March. A Survey on Predicting Heart Disease IEEE

27