ICCIEA 2020 IOP Publishing

Journal of Physics: Conference Series 1817 (2021) 012009 doi:10.1088/1742-6596/1817/1/012009

Prediction of Heart Diseases using Random Forest

Madhumita Pal1, Smita Parija2

1
Electrical engineering Dept,Govt College of Engineering, Keonjhar,
2
Electronics and Telecommunication Engg Dept, C. V Raman Global University
1
madhumitapal@gcekjr.ac.in,
2
sparija@cvrce.edu.in
Abstract. The process of discovering or mining information from a huge volume of data is known as
data mining technology. Today data mining has lots of application in every aspects of human life.
Applications of data mining are wide and diverse. Among this health care is a major application of
data mining. Medical field has get benefited more from data mining. Heart Disease is the most
dangerous life-threatening chronic disease globally. The objective of the work is to predicts the
occurrence of heart disease of a patient using random forest algorithm. The dataset was accessed from
Kaggle site. The dataset contains 303 samples and 14 attributes are taken for features of the dataset.
Then it was processed using python open access software in jupyter notebook. The datasets are
classified and processed using machine learning algorithm Random forest. The outcomes of the
dataset are expressed in terms of accuracy, sensitivity and specificity in percentage. Using random
forest algorithm, we obtained accuracy of 86.9% for prediction of heart disease with sensitivity value
90.6% and specificity value 82.7%. From the receiver operating characteristics, we obtained the
diagnosis rate for prediction of heart disease using random forest is 93.3%. The random forest
algorithm has proven to be the most efficient algorithm for classification of heart disease and therefore
it is used in the proposed system.

1. Introduction
Data mining is also known as proficiency discovering from data. It attempts to withdraw hidden
pattern and trends from huge data bases. Data mining also support automatic exploration of data. The
main objective of data mining technique is to find the hidden data in the data base. It is also called as
exploratory data analysis, data driven and deduction learning. It extracts meaningful information from
database. When the database is very large i.e in terabyte to petabytes manual analysis of data is not
possible. So, we need automatic data analysis. Data mining was introduced in 1990s.Various data
mining technologies are as follows.
(i)Statistics:
Regression analysis, cluster analysis, standard deviation etc. are the foundation of data mining.
(ii)Artificial Intelligence:
It is the applying of human thoughts like processing
(iii)Machine Learning
It is the integration of statistics and AI technology. It is about learning by the software about data.
The world is filled with data such as pictures, video, music. Machine learning promise to derive a
meaning for all the data. Arthur C. Clarke states that modern technology is filled with magic. There is
lots of data in the world generated not only from people but also from mobile, computer and from
another device. Automatic system can ascertain from data and can change the data. Machine learning
has wide application in the field of speech processing, image processing, fraud detection. Also, in the
field of medical science such as diabetes retina path, Skin cancer detection, heart disease. Using data is
referred to as for training and answer refer to as prediction. Training data refers to create a model and
to predict. This predictive model can then use to serve predictions on previously unseen data and
answer the questions. The paper is outlined as follows. Section 2 presents an idea about the related

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
ICCIEA 2020 IOP Publishing
Journal of Physics: Conference Series 1817 (2021) 012009 doi:10.1088/1742-6596/1817/1/012009

work done by the paper. Section 3 gives an idea of the major cause of heart disease, symptoms,
prevention of heart disease. Section 4 represents the result part of the experiment. Section 5 conclude
the present paper.

2. Related Work
The proposed study gives a prediction method for classification of heart disease. The risk factor which
can control and which cannot control was explained in this paper. The prediction of heart disease has
been done by random forest machine learning algorithm.
Ref [1] proposed a user-friendly heart disease prediction system (HDPS). Authors have taken 13
clinical features for classifying heart disease using artificial neural network. Prediction accuracy
obtained by the system is approximately 80%. HDPS system include clinical data section, ROC curve
section, estimation display section.
Ref [2] authors have proposed a Diabetes disease prediction system that gives diabetes malady
analysis. Two algorithms were applied namely Bayesian and K-NN for prediction of diabetes.
Ref [3] author has proposed a model for predicting heart disease by taking samples of 300 patient
record using Naïve Bayes and decision trees. data was taken from UCI repository site Author used id3
algorithm for constructing decision tree. For small data set decision tree does not give accurate result
but Naïve bayes gives more accurate result if the input data is cleaned.
Ref [4] author have proposed a data mining model to predict weather a patient has heart disease or not.
Two types of data mining algorithm decision tree and naïve bayes were used for forecasting. These
two algorithms were applied on the same data set. Decision tree show an accuracy of 91% and naïve
bayes algorithm show an accuracy of 87%. So, in the paper decision tree gives better accuracy for
predicting heart disease.
Ref [5] authors have proposed a data mining model for prediction of heart disease. Dataset was taken
from UCI machine learning repository site. Four data mining algorithms such as Naïve bayes, random
forest, Linear regression, Decision tree were applied by the authors to predict the heart disease.
Among these algorithms random forest gives good accuracy of 90.16% compared to other algorithms.
Ref [6] authors have used knn, decision tree, linear regression, support vector machine algorithms for
prediction of heart disease and compared their accuracy. All the datasets for prediction are accesses
from UCI repository site. For implementation of the algorithm’s python software is used. All the
algorithms are processed in jupyter notebook. From the experimental result authors have obtained best
accuracy of 87% by using k-nearest neighbor algorithm followed by support vector machine 83%,
decision tree 79%and linear regression of 78% accuracy among all these algorithms for prediction of
heart disease.
Ref [7] authors have proposed an application for prediction of heart disease for juveniles using
multilayer perceptron algorithms. Authors used Cleveland dataset accessed from UCI library the
dataset containing 76 parameters such as chest pain, CT scan, ECG etc. The data set was processed in
python code using PyCharm tool. From the experimental result authors obtained precision, recall,
support value for positive classes were 0.92,0.9,93and for negative classes 0.91,0.89,0.72 respectively.
Ref [8] authors have proposed a model for prediction of cardiovascular disease using machine learning
algorithm hybrid random forest with linear mode. Authors obtained 88.7% accuracy for prediction of
CVD using hybrid random forest with linear model. The dataset was collected from UCI repository
site. Authors have chosen Cleveland dataset for this proposed study.

3. Heart Disease
The Heart is the most important organ of human body. If it does not function properly then it affects
other organ of the body. According to a report 7,000,000 die from heart attacks each year. According
to WHO report around 17.9 million people die due to CVDS in 2016. 31% of the death of people is
due to Heart disease around the globe in every year. The pumping of blood to the human body is the
vital function of heart which supply oxygen and nutrients to the human body and also remove other
metabolic waste from the body. If there is deficiency of blood in human body then heart doesn’t
function properly and it stop working which causes the death of human being. Angina occurs when
there is temporary loss of blood to the heart causing chest pain. Cardiovascular disease is of two types.

2
ICCIEA 2020 IOP Publishing
Journal of Physics: Conference Series 1817 (2021) 012009 doi:10.1088/1742-6596/1817/1/012009

(1) Heart Attack-It occurs when the heart blood vessels are suddenly blocked.
(2) Heart Failure-It results from coronary heart disease, hypertension, cardiomyopathy. Heart failure is
basically when the heart is unable to maintain a strong blood flow and this results in chronic tiredness,
resist physical activities and shortness of breath. Heart failure can be divided into three types.1. right
side heart failure 2. Left side heart failure 3. congestive heart failure.
Right sided heart failure usually causes left sided heart failure. In right sided heart failure blood backs
up into other tissues such as liver and in the abdomen causing congestion in these areas. As a result of
right sided heart failure, we can have Hepatomegaly and Anciles.
In left sided heart failure oxygenated blood cannot be pumped out from heart to the rest of the body.
So, blood can back flow. Blood can accumulate in lung veins causing fluid accumulation in lungs
causes shortness of breath and oedema.

Table 1. Major cause of heart disease [10].

Disease Type
Smoking
High Blood Pressure
High Cholesterol
Diabetes and Prediabetes
Being overweight
Physical inactivity
Metabolic syndrome

Risk factor that cannot control for heart disease

1. family history
2.55 years or older
3. History of preeclampsia
Symptoms of Heart attack
Nausea
(a)Dizziness
(b)Jaw pain
(c)Abdominal pain
Living a healthy life style can reduce the effect of heart disease. Drinking plenty of water, eating green
vegetables, fat free food, doing exercises, regular check-up of heart, consulting with the doctor if there
any family history of heart disease can reduce the effect of heart disease.

4. Methodology
For the proposed study dataset was taken from Kaggle site. Then it was downloaded in excel file using
comma separated format. Data has processed by python programming using Jupiter notebook. The
data set contains 303 sample instances as shown in table3. The dataset contains 14 clinical features as
shown in table 2. Different types of python libraries such as pandas, Sklearn, NumPy, matplotlib are
used for processing the algorithms. Using explorative data analysis technique data was analysed in
jupyter notebook.10-fold cross validation technique is used for spitting the data set into training and
testing data. Then using random forest algorithm dataset was processed.
description of the algorithms:
Machine learning is the ability of computer to learn automatically from the experience.
Machine can learn by three ways.
1.supervised learning
2.Unsupervised Learning

3
ICCIEA 2020 IOP Publishing
Journal of Physics: Conference Series 1817 (2021) 012009 doi:10.1088/1742-6596/1817/1/012009

3.Reinforcement learning
In supervised learning label data is given to the machine for prediction.
K-NN, Naïve Bayes, Support vector machine, Decision tree, Random forest algorithms are supervised
machine learning algorithms.
In unsupervised learning algorithms label data is not given to the machine for prediction.
Clustering, c-means are the examples of unsupervised learning
In reinforcement learning machine learn by itself without any guidance. It learns from the environment
and there is a reward for every action.
Q-learning is one of the examples of Reinforcement learning.
Random forest is a supervised machine learning algorithm that constructs several decision trees. The
final decision is made based on the majority of decision tree. Decision tree suffer from low bias and
high variance. Random forest converts high variance to low variance.
Table 2. Features for data prediction
Attribute meaning
Age1 Age is continuous
Gender 1 1=male 0=female
Cp1 Chest pain

Trestbps Resting blood pressure results during hospitalised:

continuous(mmHg)
chol cholesterol level in mg/dl
Fbs1 Fasting blood sugar
0:<=120mg/dl,1:>120mg/dl
restecg electrocardiographic results during resting 1=true 0=false
thalach Maximum heart rate achieved: continuous
exang Exercise induced angina

oldpeak ST depression
slope ST segment slope

ca Number of major vessels coloured by fluoroscopy: discrete

(0,1,2,3)
thal 3: normal
6: fixed defect
7: reversible defect

5. Result and Discussion

The present work predicts suffering rate of a patient from heart disease using random forest
algorithm. Total 303 data samples (Table 3) of 14 clinical features (Table 2) have taken for prediction
of heart disease.80% of the dataset has taken for training and 20% has taken for testing phase.

Table 3. Features for data prediction

4
ICCIEA 2020 IOP Publishing
Journal of Physics: Conference Series 1817 (2021) 012009 doi:10.1088/1742-6596/1817/1/012009

Figure 1. correlation matrix of the dataset

After doing explorative data analysis we obtained the correlation matrix which correlate the attributes
of the data set.
We are applying random forest algorithm to the testing data set for creating a confusion matrix. From
the confusion matrix we get more sophisticated metrics like sensitivity, specificity and AUC that can
help us to make a decision in the classification process.

5
ICCIEA 2020 IOP Publishing
Journal of Physics: Conference Series 1817 (2021) 012009 doi:10.1088/1742-6596/1817/1/012009

Figure 2. Confusion matrix obtained in the experimental work

From the confusion matrix we get 24 correctly classified negative class and 29 correctly classified
positive class. Three incorrectly classified negative class and five correctly classified positive class as
shown in table 4.

Table 4. Result of Confusion matrix

True positive 29
True negative 24
False positive 5
False negative 3
Sensitivity 90.6
Specificity 82.7
Accuracy 86.9

From Table 4 we obtained sensitivity value as 90.6% that tells us 90.6% of patients with heart disease
were correctly classified. Similarly, we obtained the specificity value as 82.7% that tells us 82.7% of
patients without heart disease were correctly classified. So, from the experiment we get that random
forest correctly predicts 29 classes of patients with heart disease and 24 classes of patients without
heart disease.

6
ICCIEA 2020 IOP Publishing
Journal of Physics: Conference Series 1817 (2021) 012009 doi:10.1088/1742-6596/1817/1/012009

Figure 3.ROC curve obtained using random forest algorithm

The ROC curve between true positive rate and false positive rate at different threshold level is plotted.
From the ROC curve we obtained the AUC value is 93.3% that indicates the model 93.3% accurately
predict whether the patient suffered from heart disease or not.

6. Conclusion

In this paper random forest data mining algorithm was implemented for prediction of heart disease.
From the experimental work we obtained the Sensitivity value 90.6%. specificity value 82.7, and
accuracy value of 86.9 for prediction. In the proposed work we obtained classification accuracy of
86.9%for prediction of heart disease with diagnosis rate of 93.3% using random forest algorithm. The
proposed system can also be used for prediction of other disease by applying with other machine
learning algorithm such as Naïve Bayes, decision tree, K-NN, Linear regression, fuzzy logic for better
accuracy. Cloud computing technology can also be used for the proposed system to manage large
volume of patient data.

References

[1] Chen, A. H., Huang, S. Y., Hong, P. S., Cheng, C. H., & Lin, E. J. (2011, September). HDPS:
“Heart disease prediction system”. In 2011 Computing in Cardiology (pp. 557-560). IEEE.

[2] Shetty, Deeraj, Kishor Rit, Sohail Shaikh, and Nikita Patil. "Diabetes disease prediction using
data mining."In 2017 International Conference on Innovations in Information, Embedded and
Communication Systems (ICIIECS), pp. 1-5. IEEE, 2017.
[3] Rajesh , T Maneesha, Shaik Hafeez, Hari Krishna“Prediction of Heart Disease Using Machine
Learning Algorithms“May 2018International Journal of Engineering & Technology 7(2):363-366DOI:
10.14419/ijet. v7i2.32.15714
North-Holland/American Elsevier) p 517
[4] J. Krishnan Santana; S. Geetha “Prediction of Heart Disease Using Machine Learning
Algorithms”. 2019 1st International Conference on Innovations in Information and Communication
Technology (ICIICT)Publisher: IEEE

[5] Rajdhan Apurb, Agarwal Avi, Sai Milan, Ravi Dundigalla, Ghuli Poonam.” Heart Disease
Prediction using Machine Learning” INTERNATIONAL JOURNAL OF ENGINEERING
RESEARCH & TECHNOLOGY

7
ICCIEA 2020 IOP Publishing
Journal of Physics: Conference Series 1817 (2021) 012009 doi:10.1088/1742-6596/1817/1/012009

[6] Singh, A., & Kumar, R. (2020). “Heart Disease Prediction Using Machine Learning
Algorithms”. 2020 International Conference on Electrical and Electronics Engineering (ICE3).
doi:10.1109/ice348803.2020.9122958
[7] Mohan, S., Thirumalai, C., & Srivastava, G. (2019). “Effective Heart Disease Prediction using
Hybrid Machine Learning Techniques”. IEEE Access, 1–1. doi:10.1109/access.2019.2923707
[8] Rairikar, A., Kulkarni, V., Sabale, V., Kale, H., & Lamgunde, A. (2017, June). “Heart disease
prediction using data mining techniques”. In 2017 International Conference on Intelligent Computing
and Control(I2C2) (pp. 1-8). IEEE
[9] Al Essa, Ali Radhi, and Christian Bach. "Data Miningand Warehousing." American Society for
EngineeringEducation (ASEE Zone 1) Journal (2014).
[10] National Health Council, ‘Heart HealthScreenings’,2017. [Online]Available:
http://www.heart.org/HEARTORG/Conditions/HeartHealthScreenings_UCM_428687_Article.jsp#.
WnsOAeeYPIV