Heart Disease Prediction using Machine Learning

Algorithms

By
Bishal Ghosh
Roll no: 25500115014
Registration no: 152550110015
Chayan Datta
Roll no: 25500115015
Registration no: 152550110016
Nilotpal Majumdar
Roll no: 25500115024
Registration no: 1525500110026
Sayan Paul
Roll no: 25500115038
Registration no: 152550110040

Under the Guidance of

Mrs. Debdatta Chatterjee

Bachelor of Technology(Computer Science & Engineering)

Department of Computer Science & Engineering
Dr. Sudhir Chandra Sur Degree Engineering College
Maulana Abul Kalam Azad Institute of Technology
Kolkata, West Bengal
 CERTIFICATE

This is to certify that this is a bona fide record to the FINAL YEAR project work “Heart Disease
Prediction using Machine Learning” done satisfactorily at DR. SUDHIR CHANDRA SUR
DEGREE ENGINEERING COLLEGE by BISHAL GHOSH, CHAYAN DATTA, NILOTPAL
MAJUMDAR, SAYAN PAUL of 8th SEMESTER, CSE.

This report or similar report on this topic has not been submitted for any other examination and
does not form part of any other course undergone by the candidate. I have no doubt that they
have a very good research potential.

I would like to wish them a very bright future.

Date: ____________ ___________________

Mrs. Debdatta Chatterjee

(Prof, CSE, DSCSDEC)

___________________ ____________________

Mrs. Mallika De (EXAMINER)

(HOD,CSE,DSCSDEC)
 ACKNOWLEDGEMENT

We would like to take this opportunity to express our gratitude towards all those people who
have in various ways, helped in the successful completion of our final year project work on
“Heart Disease Prediction using Machine Learning Algorithms” done satisfactorily at DR.
SUDHIR CHANDRA SUR DEGREE ENGINEERING COLLEGE. We must convey our
gratitude to Mrs. Debdatta Chatterjee and Mr. Sovan Saha for giving us the constant source
of inspiration and help in preparing the project, personally correcting our work and providing
encouragement throughout the project. In that regard we are especially thankful to our Head of
the Department Mrs. Mallika De for steering me through the tough as well as easy phases of the
project in a result-oriented manner with concern attention. A special thanks to all the faculty
members of our CSE department.

We would also love to take this opportunity to tell the readers of our material that their
comments, be it appreciation or criticism would be the most valuable thing to us. That will be
something, we will always be thankful for.

Date: ________________
 ABSTRACT

Nowadays, health disease are increasing day by day due to lifestyle, hereditary. Especially, heart
disease has become more common these days, i.e. life of people is at risk. Each individual has
different values for Blood pressure, cholesterol and pulse rate. But according to medically proven
results the normal values of Blood pressure is 120/90, Cholesterol is 100-129 mg/dL ,Pulse rate is
72, Fasting Blood Sugar level is 100 mg/dL ,Heart rate is 60-100 bpm, ECG is normal, Width of
major vessels is 25 mm (1 inch) in the aorta to only 8 μm in the capillaries. This paper gives the
survey about different classification techniques used for predicting the risk level of each person
based on age, gender, Blood pressure, cholesterol, pulse rate.

“Disease Prediction” system based on predictive modeling predicts the disease of the user
on the basis of the symptoms that user provides as an input to the system. The system analyzes the
symptoms provided by the user as input and gives the probability of the disease as an output.
Disease Prediction is done by implementing 5 techniques such as Naïve Bayes, KNN, Decision
Tree, Linear Regression and Random Forest Algorithms. These techniques calculate the
probability of the disease. Therefore, average prediction accuracy probability 83% is obtained.
 CONTENTS

CHAPTER 1: INTRODUCTION

1.1 Introduction 1

1.2 Motivation 4

1.3 Literature Review 5

1.4 What is Heart Disease? 10

1.4.1 How the Heart works? 11

1.4.2 Chambers and valves of the heart 11

1.4.3 Heart valves 12

1.4.4 Causes of cardiovascular disease 13

1.4.5 Risk factors 15

1.4.6 Complications 16

1.4.7 Prevention 17

1.5 Some of the attributes we used for Heart Disease Prediction

and their correlation to CVD 17

1.5.1 Age as a Cardiovascular Risk Factor 18

1.5.2 Gender differences in coronary heart disease 22

1.5.3 The association between blood pressure and mortality

in patients with heart failure 28

1.5.4 Chest Pain and its risk factor to Cardiac arrest 30

1.5.5 Cholesterol and Heart Disease 32

 1.5.6 Fasting Glucose Level or Fasting Blood Sugar

and the Risk of Heart Diseases 33

1.5.7 Electrocardiograph (ECG) Test for Heart Diseases 35

1.5.8 Cardiac Complications in Thalassemia Major 36

CHAPTER 2: RELATED WORKS

2.1 Prediction system for heart disease using Naive Bayes and particle swarm 40

2.1.1 Particle swarm optimization (PSO) 40

2.1.2 Naïve Bayes’ Classifier 40

2.2 Predictive Data Mining for Medical Diagnosis: Heart Disease Prediction 41

2.2.1 Data Mining in the Heart Disease Prediction 42

2.2.2 Data Mining and Artificial Neural Network 43

2.2.3 Data Mining and Genetic Algorithm 44

2.2.4 Association Rule Discovery 45

2.2.5 Issues and Challenges 46

CHAPTER 3: PRESENT WORK

3.1 About Heart Disease 47

3.2 Motivation 48

3.3 Proposal 48

3.4 Dataset Structure & Description 49

3.4.1 Importing libraries 49

 3.4.2 Load Data 50

3.4.3 Check the type of the dataset 50

3.4.4 Check the Shape of the data 50

3.4.5 Check the top four columns of the dataset 50

3.4.6 Dataset description 51

3.4.7 Types of features 52

3.4.8 Some Random data columns 53

3.4.9 Check for missing Data 53

3.4.10 Check the correlation with target data 54

3.5 Exploratory Data Analysis (EDA) 54

3.5.1 Percentage of patient with or without heart problems in the given dataset 55

3.5.2 Uniqueness of sex column 56

3.5.3 Check the percentage and plot the graph 56

3.5.4 Heart Disease Frequency for Ages 57

3.5.5 Heart Disease Frequency for sex 59

3.5.6 Making hr data column names easily recognizable 60

3.5.7 Checking out Male/Female Heart disease according to

Fasting Blood Sugar 60

3.5.8 Analyzing the chest pain 62

3.5.9 Analyzing the resting blood pressure 63

3.5.10 Analyzing the resting electrocardiographic measurement 66

3.5.11 Analyzing Exercise Induced angina 67

3.5.12 Slope of the peak exercise ST segment 68

 3.5.13 Analyzing no. of major vessels coloured by fluoroscopy 72

3.5.14 Analyzing thalassemia 76

3.5.15 Thalassemia vs cholesterol 78

3.6 Correlation Matrix 80

3.7 Data Preparation 80

3.8 Modelling and predicting with Machine Learning 81

3.8.1 Logistic Regression 84

3.8.2 Random Forest 85

3.8.3 Naïve Bayes 88

3.8.4 K-Nearest Neighbour 91

3.8.5 Decision Tree 94

3.9 Result 98

3.10 System Requirements 99

CHAPTER: 4 CONCLUSIONS

4.1 Inference 100

4.2 Drawbacks 102

4.3 Future Scope 102

REFERENCES 104
 Heart Disease Prediction using Machine Learning Algorithms

CHAPTER: 1 INTRODUCTION

1.1 Introduction

In day to day life many factors that affect a human heart. Many problems are occurring at a
rapid pace and new heart diseases are rapidly being identified. In today’s world of stress Heart,
being an essential organ in a human body which pumps blood through the body for the blood
circulation is essential and its health is to be conserved for a healthy living. The health of a
human heart is based on the experiences in a person’s life and is completely dependent on
professional and personal behaviors of a person. There may also be several genetic factors
through which a type of heart disease is passed down from generations. According to the World
Health Organization, every year more than 12 million deaths are occurring worldwide due to
the various types of heart diseases which is also known by the term cardiovascular disease. The
term Heart disease includes many diseases that are diverse and specifically affect the heart and
the arteries of a human being. Even young aged people around their 20-30 years of lifespan are
getting affected by heart diseases. The increase in the possibility of heart disease among young
may be due to the bad eating habits, lack of sleep, restless nature, depression and numerous
other factors such as obesity, poor diet, family history, high blood pressure, high blood
cholesterol, idle behavior, family history, smoking and hypertension.

The diagnosis of the heart diseases is a very important and is itself the most complicated task
in the medical field. All the mentioned factors are taken into consideration when analyzing and
understanding the patients by the doctor through manual check-ups at regular intervals of time.

The symptoms of heart disease greatly depend upon which of the discomfort felt by an
individual. Some symptoms are not usually identified by the common people. However,
common symptoms include chest pain, breathlessness, and heart palpitations. The chest pain
common to many types of heart disease is known as angina, or angina pectoris, and occurs
when a part of the heart does not receive enough oxygen. Angina may be triggered by stressful
events or physical exertion and normally lasts under 10 minutes. Heart attacks can also occur

1
 Heart Disease Prediction using Machine Learning Algorithms

as a result of different types of heart disease. The signs of a heart attack are like angina except
that they can occur during rest and tend to be more severe. The symptoms of a heart attack can
sometimes resemble indigestion. Heartburn and a stomach ache can occur, as well as a heavy
feeling in the chest. Other symptoms of a heart attack include pain that travels through the
body, for example from the chest to the arms, neck, back, abdomen, or jaw, lightheadedness
and dizzy sensations, profuse sweating, nausea and vomiting.

Heart failure is also an outcome of heart disease, and breathlessness can occur when the heart
becomes too weak to circulate blood. Some heart conditions occur with no symptoms at all,
especially in older adults and individuals with diabetes. The term 'congenital heart disease'
covers a range of conditions, but the general symptoms include sweating, high levels of fatigue,
fast heartbeat and breathing, breathlessness, chest pain. However, these symptoms might not
develop until a person is older than 13 years. In these types of cases, the diagnosis becomes an
intricate task requiring great experience and high skill. A risk of a heart attack or the possibility
of the heart disease if identified early, can help the patients take precautions and take regulatory
measures. Recently, the healthcare industry has been generating huge amounts of data about
patients and their disease diagnosis reports are being especially taken for the prediction of heart
attacks worldwide. When the data about heart disease is huge, the machine learning techniques
can be implemented for the analysis.

Data Mining is a task of extracting the vital decision-making information from a collective of
past records for future analysis or prediction. The information may be hidden and is not
identifiable without the use of data mining. The classification is one data mining technique
through which the future outcome or predictions can be made based on the historical data that
is available. The medical data mining made a possible solution to integrate the classification
techniques and provide computerized training on the dataset that further leads to exploring the
hidden patterns in the medical data sets which is used for the prediction of the patient’s future
state. Thus, by using medical data mining it is possible to provide insights on a patient’s history
and is able to provide clinical support through the analysis. For clinical analysis of the patients,
these patterns are very much essential. In simple English, the medical data mining uses
classification algorithms that is a vital part for identifying the possibility of heart attack before
the occurrence. The classification algorithms can be trained and tested to make the predictions
that determine the person’s nature of being affected by heart disease[1].

2
 Heart Disease Prediction using Machine Learning Algorithms

In this research work, the supervised machine learning concept is utilized for making the
predictions. A comparative analysis of the three data mining classification algorithms namely
Random Forest, Decision Tree and Naïve Bayes are used to make predictions. The analysis is
done at several levels of cross validation and several percentage of percentage split evaluation
methods respectively. The Stat Log dataset from UCI machine learning repository is utilized
for making heart disease predictions in this research work. The predictions are made using the
classification model that is built from the classification algorithms when the heart disease
dataset is used for training. This final model can be used for prediction of any types of heart
diseases.[2][3]

Health-care is a field of the most needed service and an economically 2ndlargest industry in
21stcentury. While we talk about the affordability and quality assurance in health-care industry,
several statistical analyses are carried on making health solutions more precise and flawless in
this current era of increasing health problems and chronic diseases. Advancements on data
driven intelligent technologies is disease diagnosis and detection, treatment and research are
remarkable. Medical image analysis, symptom-based disease prediction is the part where the
most sought-after brains are working. In this paper we aim to present our proposed model on
the prediction on diagnosis of cardio vascular disease with ECG analysis and symptom-based
detection. The model aims to be researched and advance in further to become robust and end
to end reliable research tool. We will discuss about the classical methods and algorithms
implemented on CVD prediction, gradual advancements, draw comparison of performance
among the existing systems and propose an enhanced multi-module system performing better
in terms of accuracy and feasibility. Implementation, training and testing of the modules have
been done on datasets obtained from UCI and Physio net data repositories. Data format have
been modified in case of the ECG report data for betterment of action by the convolutional
neural network used in our research and in the risk prediction module we have chosen attributes
for training and implementing the multi-layered neural network developed by us. The further
research and advancement possibilities are also mentioned in the paper. [4]

3
 Heart Disease Prediction using Machine Learning Algorithms

1.2 Motivation

The main motivation of doing this research is to present a heart disease prediction model for
the prediction of occurrence of heart disease. Further, this research work is aimed towards
identifying the best classification algorithm for identifying the possibility of heart disease in a
patient. This work is justified by performing a comparative study and analysis using three
classification algorithms namely Naïve Bayes, Decision Tree, KNN, Logistic Regression and
Random Forest are used at different levels of evaluations. Although these are commonly used
machine learning algorithms, the heart disease prediction is a vital task involving highest
possible accuracy. Hence, the three algorithms are evaluated at numerous levels and types of
evaluation strategies. This will provide researchers and medical practitioners to establish a
better understanding and help them identify a solution to identify the best method for predicting
the heart diseases[5][6].

A key challenge confronting healthcare organization (hospitals, medical centers) is the

facility of quality services at reasonable prices. Quality amenities suggest diagnosing patients
accurately and regulating medications that are effective. Poor clinical choices can prompt
deplorable results, which are in this manner unsatisfactory. Hospitals should limit the cost of
clinical tests. They can accomplish these outcomes by utilizing fitting PC based data and
additionally choice emotionally supportive networks [7][8]. The heart is the essential piece of
our body. Life is itself reliant on effective working of the heart. If task of the heart isn't
legitimate, it will influence the other body parts of human, for example, cerebrum, kidney and
so on. Coronary illness is a sickness that effects on the activity of the heart. There are several
elements which builds danger of Heart ailment [9].

Some of them are listed below:

• The family history of heart disease

• Smoking

• Cholesterol

• High blood pressure

• Obesity

4
 Heart Disease Prediction using Machine Learning Algorithms

• Lack of physical exercise

Because of a wide accessibility of superlative measure of information and a need to change

over this accessible huge measure of information to helpful data requires the utilization of
information mining strategies. Information Mining and KDD (learning disclosure in the
database) have turned out to be prominent as of late[10][11]. The popularity of information
mining and KDD (information revelation in database) shouldn't be an amazement since the
measure of the information increases that are accessible are extremely extensive to be
analyzed physically and even the techniques for programmed information investigation in
view of established insights and machine adapting frequently threaten issues when preparing
large, dynamic information increases comprising of complex items [12]. Information Mining
is the center piece of Knowledge Discovery Database (KDD). Numerous individuals regard
Data Mining as an equivalent word for KDD since it's a key piece of KDD process[13][14].
There are sure stages of information mining that you will need to get comfortable with, and
these are exploration, pattern identification, and deployment. Information mining is an
iterative procedure that commonly includes the accompanying stage [15].

1.3 Literature Review

According to Ordonez [16]the heart disease can be predicted with some basic attributes taken
from the patient and in their work have introduced a system that includes the characteristics of
an individual human being based on totally 13 basic attributes like sex, blood pressure,
cholesterol and others to predict the likelihood of a patient getting affected by heart disease.
They have added two more attributes i.e. fat and smoking behavior and extended the research
dataset. The data mining classification algorithms such as Decision Tree, Naive Bayes, and
Neural Network are utilized to make predictions and the results are analyzed on Heart disease
database.

Yilmaz, [17]have proposed a method that uses least squares support vector machine (LS-SVM)
utilizing a binary decision tree for classification of cardiotocogram to find out the patient
condition.

5
 Heart Disease Prediction using Machine Learning Algorithms

Duff, et al. [18]have done a research work involving five hundred and thirty-three patients who
had suffered from cardiac arrest and they were integrated in the analysis of heart disease
probabilities. They performed classical statistical analysis and data mining analysis using
mostly Bayesian networks.

Frawley, et al. [19] have performed a work on prediction of survival of Coronary heart disease
(CHD) which is a challenging research problem for medical society. They also used 10-fold
cross-validation methods to determine the impartial estimate of the three prediction models for
performance comparison purposes.

Lee et al.[20] proposed a novel methodology to expand and study the multi-parametric feature
along with linear and nonlinear features of Heart Rate Variability diagnosing cardiovascular
disease. They have carried out various experiments on linear and non-linear features to estimate
several classifiers, e.g., Bayesian classifiers, CMAR, C4.5 and SVM. Based on their
experiments, SVM outperformed the other classifiers.

Noh et al. [21]suggested a classification method which is an associative classifier that is

constructed based on the efficient FP-growth method. Because the volume of patterns can be
diverse and huge, they offered a rule to measure the cohesion and in turn allow a tough choice
of pruning patterns in the pattern-generating process.

Parthiban, et al. [22] have proposed a new work in which the heart disease is identified and
predicted using the proposed Coactive Neuro-Fuzzy Inference System (CANFIS). Their model
works based on the collective nature of neural network adaptive capabilities and based on the
genetic algorithm along with fuzzy logic in order to diagnose the occurrence of the disease.
The performance of the proposed CANFIS model was evaluated in terms of training
performances and classification accuracies. Finally, their results show that the proposed
CANFIS model has great prospective in predicting the heart disease.

Singh, et al. [23] have done a work using, one partition clustering algorithm (K-Means) and
one hierarchical clustering algorithm (agglomerative). K-means algorithm has higher
effectiveness and scalability and converges fast when production with large data sets.
Hierarchical clustering constructs a hierarchy of clusters by either frequently merging two
smaller clusters into a larger one or splitting a larger cluster into smaller ones. Using WEKA
data mining tool, they have calculated the performance of k-means and hierarchical clustering
algorithm on the basis of accuracy and running time.

6
 Heart Disease Prediction using Machine Learning Algorithms

Guru, et al. [24] have proposed the computational model based on a multilayer perceptron with
three layers is employed to enlarge a decision support system for the finding of five major heart
diseases. The proposed decision support system is trained using a back propagation algorithm
amplified with the momentum term, the adaptive learning rate and the forgetting mechanics.

Palaniappan, et al. [25] have carried out a research work and have built a model known as
Intelligent Heart Disease Prediction System (IHDPS) by using several data mining techniques
such as Decision Trees, Naïve Bayes and Neural Network.

Shantakumar, et al. [26] have done a research work in which the intelligent and effective heart
attack prediction system is developed using Multi-Layer Perceptron with Back-Propagation.
Accordingly, the frequency patterns of the heart disease are mined with the MAFIA algorithm
based on the data extracted.

Yanwei, et.al [27] have built a classification method based on the origin of multi parametric
features by assessing HRV (Heart Rate Variability) from ECG and the data is pre-processed
and heart disease prediction model is built that classifies the heart disease of a patient.

Data mining plays an important role in the field of heart disease prediction. [28] Medical Data
mining has great potential like exploring the hidden patterns which can be utilized for clinical
diagnosis of any disease dataset[29]. Several data mining techniques are used in the diagnosis
of heart disease such as Naive Bayes, Decision Tree, neural network, kernel density, bagging
algorithm, and support vector machine showing different levels of accuracies. Naive Bayes is
one of the successful classification techniques used in the diagnosis of heart disease patients.
Peter et al. [30] talked about a new feature selection method algorithm which is the hybrid
method which combined CFS and Bayes theorem (CFS+Filter Subset Eval) and evaluated
accuracy 85.5%.

Shouman [31] presented work by integrating k-means clustering with Naive Bayes using
different initial centroid selection to improve the Naive Bayes accuracy for diagnosing heart
disease patients and accuracy was 84.5%. Rupali et al. [32] decision support in Heart Disease
Prediction

System (HDPS) is developed by using both Naive Bayesian Classification and Jelinek-Mercer
smoothing technique. This Laplace smoothing is used to make an approximating function
which attempts to capture important patterns in the data to avoid noise & accuracy is 86%.
Elma et al. [33] proposed a classifier with the distance-based algorithm K-nearest neighbor and

7
 Heart Disease Prediction using Machine Learning Algorithms

statistical based Naïve Bayes classifier (cNK) and achieved the accuracy 85.92% for heart
disease dataset.[34]

Data mining has been played an important role in the intelligent medical systems [35][36]. The
relationships of disorders and the real causes of the disorders and the effects of symptoms that
are spontaneously seen in patients can be evaluated by the users via the constructed software
easily. Large databases can be applied as the input data to the software by using the
extendibility of the software. The effects of relationships that have not been evaluated
adequately have been explored and the relationships of hidden knowledge laid among the large
medical databases have been searched in this study by means of finding frequent items using
candidate generation. The sets of sicknesses simultaneously seen in the medical databases can
be reduced by using our non-candidate approach. Knowledge of the risk factors associated with
heart disease helps health care professionals to identify patients at high risk of having heart
disease. Statistical analysis and data mining techniques to help healthcare professionals in the
diagnosis of heart disease. Statistical analysis has identified the disorders of the heart and blood
vessels, and includes coronary heart disease (heart attacks), cerebrovascular disease (stroke),
raised blood pressure (hypertension), peripheral artery disease, rheumatic heart disease,
congenital heart disease and heart failure. The major causes of cardiovascular disease are
tobacco use, physical inactivity, an unhealthy diet and harmful use of alcohol. The three major
causes of heart diseases are chest pain, stroke and heart attack [37]. The data mining methods
like artificial neural network technique is used in effective heart attack prediction system. First
the dataset used for prediction of heart diseases was pre-processed and clustered by means of
K-means clustering algorithm [38]. Then neural network is trained with the selected significant
patterns. Multi-layer Perceptron Neural Network with Back-propagation is used for training.
The results indicate that the algorithm used is capable of predicting the heart diseases more
efficiently. The prediction of heart diseases significantly uses 15 attributes, with basic data
mining technique like ANN, Clustering and Association Rules, soft computing approaches etc.
The outcome shows that Decision Tree performance is more and few time Bayesian
classification is having similar accuracy as of decision tree but other predictive methods like
K-Nearest Neighbor, Neural Networks, Classification based on clustering will not perform well
[39]. By using the Weighted Associative Classifier (WAC), a slight change has been made,
instead of considering 5 class labels, only 2 class labels are used. One for “Heart Disease” and
another one for “No Heart Disease”. The maximum accuracy (81.51%) has been achieved.

8
 Heart Disease Prediction using Machine Learning Algorithms

When genetic algorithm is applied, the accuracy of the Decision Tree and Bayesian
Classification is improved by reducing the actual data size. The dataset of 909 patient records
with heart diseases has been collected and 13 attributes has been used for consistency [37]. The
patient records have been splatted equally as 455 records for training dataset and 454 records
for testing dataset. After applying genetic algorithm, the attributes has been reduced to 6 and
decision tree performs more efficiently with 99.2% accuracy when compared with other
algorithms. In 2011, Hnin Wint Khaing presented an efficient approach for the prediction of
heart attack risk levels from the heart disease database. Firstly, the heart disease database is
clustered using the K-means clustering algorithm, which will extract the data relevant to heart
attack from the database. This approach allows mastering the number of fragments through its
k parameter. Subsequently the frequent patterns are mined from the extracted data, relevant to
heart disease, using the MAFIA (Maximal Frequent Item set Algorithm) algorithm. The
machine learning algorithm is trained with the selected significant patterns for the effective
prediction of heart attack. They have employed the ID3 algorithm as the training algorithm to
show level of heart attack with the decision tree. The results showed that the designed
prediction system is capable of predicting the heart attack effectively [40]. Chourasia and Pal
conducted study on the prediction of heart attack risk levels from the heart disease data base.
The prediction of heart diseases significantly uses 11important attributes, with basic data
mining technique like Naïve Bayes, J48 decision tree and Bagging approaches. The outcome
shows that bagging techniques performance is more accurate than Bayesian classification
andJ48.The results shows that the bagging prediction system is capable of predicting the heart
attack effectively[41]. Researchers have been applying various algorithms and techniques like
Classification, Clustering, Regression, Artificial Intelligence, Neural Networks, Association
Rules, Decision Trees, Genetic Algorithm, Nearest Neighbor method etc., to help health care
professionals with improved accuracy in the diagnosis of heart disease. The heart disease
database used from the University of California Irvine. UCI archive is used. This database
contains four data sets from the Cleveland Clinic Foundation, Hungarian Institute of
Cardiology, V.A. Medical Center and University Hospital of Switzerland. However, here we
discuss the Cleveland Heart Disease Dataset (CHDD). Data Mining Techniques. This paper
uses data mining algorithms CART (Classification and Regression Tree), ID3 (Iterative
Dichotomized 3) and decision table (DT). These classification algorithms are selected because
they are very often used for research purposes and have potential to yield good results.

9
 Heart Disease Prediction using Machine Learning Algorithms

Moreover, they use different approaches for generating the classification models, which
increases the chances for finding a prediction model with high classification accuracy. CART.

1.4 What is Heart Disease?

Heart disease describes a range of conditions that affect your heart. Diseases under the heart
disease umbrella include blood vessel diseases, such as coronary artery disease; heart rhythm
problems (arrhythmias); and heart defects you're born with (congenital heart defects), among
others. The term "heart disease" is often used interchangeably with the term "cardiovascular
disease." Cardiovascular disease generally refers to conditions that involve narrowed or
blocked blood vessels that can lead to a heart attack, chest pain (angina) or stroke. Other heart
conditions, such as those that affect your heart's muscle, valves or rhythm, also are considered
forms of heart disease. Heart failure is a serious condition with high prevalence (about2% in
the adult population in developed countries, and or than8%inpatients olderthan75years). About
3 – 5%ofhospitaladmissions are linked with heart failure incidents. Heart failure is the first
cause of admission by healthcare professionals in their clinical practice. The costs are very
high, reaching up to 2% of the total health costs in the developed countries. Building an
effective disease management strategy requires analysis of large amount of data, early detection
of the disease, assessment of the severity and early prediction of adverse events. This will
inhibit the progression of the disease, will improve the quality of life of the patients and will
reduce the associated medical costs. Toward this direction machine learning techniques have
been employed. The aim of this paper is to present the state-of-the-art of the machine learning
methodologies applied for the assessment of heart failure. More specifically, models predicting
the presence, estimating the subtype, assessing the severity of heart failure and predicting the
presence of adverse events, such as destabilizations, re-hospitalizations, and mortality are
presented. According to the authors' knowledge, it is the first time that such a comprehensive
review, focusing on all aspects of the management of heart failure, is presented.

10
 Heart Disease Prediction using Machine Learning Algorithms

1.4.1 How the Heart Works?




Figure 1 Working of the Heart

1.4.2 Chambers and valves of the heart

Your heart is a pump. It's a muscular organ about the size of your fist, situated slightly left of
center in your chest. Your heart is divided into the right and the left side. The division prevents
oxygen-rich blood from mixing with oxygen-poor blood. Oxygen-poor blood returns to the
heart after circulating through your body.

11
 Heart Disease Prediction using Machine Learning Algorithms

 The right side of the heart, comprising the right atrium and ventricle, collects and pumps
blood to the lungs through the pulmonary arteries.

 The lungs refresh the blood with a new supply of oxygen. The lungs also breathe out
carbon dioxide, a waste product.

 Oxygen-rich blood then enters the left side of the heart, comprising the left atrium and
ventricle.

 The left side of the heart pumps blood through the aorta to supply tissues throughout
the body with oxygen and nutrients.

1.4.3 Heart valves

Four valves within your heart keep your blood moving the right way by opening only one way
and only when they need to. To function properly, the valve must be formed properly, must
open all the way and must close tightly so there's no leakage. The four valves are:

 Tricuspid

 Mitral

 Pulmonary

 Aortic

12
 Heart Disease Prediction using Machine Learning Algorithms

1.4.4 Causes of cardiovascular disease

Figure 2 Causes of Cardiovascular Diseases

Development of atherosclerosis - While cardiovascular disease can refer to different heart or

blood vessel problems, the term is often used to mean damage to your heart or blood vessels
by atherosclerosis (ath-ur-o-skluh-ROE-sis), a buildup of fatty plaques in your arteries. Plaque
buildup thickens and stiffens artery walls, which can inhibit blood flow through your arteries
to your organs and tissues. Atherosclerosis is also the most common cause of cardiovascular
disease. It can be caused by correctable problems, such as an unhealthy diet, lack of exercise,
being overweight and smoking.

Causes of heart arrhythmia - Common causes of abnormal heart rhythms (arrhythmias) or

conditions that can lead to arrhythmias include:

13
 Heart Disease Prediction using Machine Learning Algorithms

 Heart defects you're born with (congenital heart defects)

 Coronary artery disease
 High blood pressure
 Diabetes
 Smoking
 Excessive use of alcohol or caffeine
 Drug abuse
 Stress
 Some over-the-counter medications, prescription medications, dietary supplements and herbal
remedies
 Valvular heart disease

In a healthy person with a normal, healthy heart, it's unlikely for a fatal arrhythmia to develop
without some outside trigger, such as an electrical shock or the use of illegal drugs. That's
primarily because a healthy person's heart is free from any abnormal conditions that cause an
arrhythmia, such as an area of scarred tissue. However, in a heart that's diseased or deformed,
the heart's electrical impulses may not properly start or travel through the heart, making
arrhythmias more likely to develop.

Causes of congenital heart defects - Congenital heart defects usually develop while a baby is
in the womb. Heart defects can develop as the heart develops, about a month after conception,
changing the flow of blood in the heart. Some medical conditions, medications and genes may
play a role in causing heart defects. Heart defects can also develop in adults. As you age, your
heart's structure can change, causing a heart defect.

Causes of cardiomyopathy - The cause of cardiomyopathy, a thickening or enlarging of the

heart muscle, may depend on the type:

 Dilated cardiomyopathy. The cause of this most common type of cardiomyopathy often is
unknown. It may be caused by reduced blood flow to the heart (ischemic heart disease)
resulting from damage after a heart attack, infections, toxins and certain drugs. It may also be
inherited from a parent. It usually enlarges (dilates) the left ventricle.
 Hypertrophic cardiomyopathy. This type, in which the heart muscle becomes abnormally
thick, usually is inherited. It can also develop over time because of high blood pressure or
aging.
 Restrictive cardiomyopathy. This least common type of cardiomyopathy, which causes the
heart muscle to become rigid and less elastic, can occur for no known reason. Or it may be
caused by diseases, such as connective tissue disorders, excessive iron buildup in your body

14
 Heart Disease Prediction using Machine Learning Algorithms

(hemochromatosis), the buildup of abnormal proteins (amyloidosis) or by some cancer

treatments.

Causes of heart infection - A heart infection, such as endocarditis, is caused when an irritant,
such as a bacterium, virus or chemical, reaches your heart muscle. The most common causes
of heart infection include:

 Bacteria
 Viruses
 Parasites

Causes of valvular heart disease - There are many causes of diseases of your heart valves.
You may be born with valvular disease, or the valves may be damaged by conditions such as:

 Rheumatic fever
 Infections (infectious endocarditis)
 Connective tissue disorders

1.4.5 Risk factors

Risk factors for developing heart disease include

 Age. Aging increases your risk of damaged and narrowed arteries and weakened or
thickened heart muscle.
 Sex. Men are generally at greater risk of heart disease. However, women's risk
increases after menopause.
 Family history. A family history of heart disease increases your risk of coronary
artery disease, especially if a parent developed it at an early age (before age 55 for a
male relative, such as your brother or father, and 65 for a female relative, such as your
mother or sister).
 Smoking. Nicotine constricts your blood vessels, and carbon monoxide can damage
their inner lining, making them more susceptible to atherosclerosis. Heart attacks are
more common in smokers than in nonsmokers.
 Certain chemotherapy drugs and radiation therapy for cancer. Some
chemotherapy drugs and radiation therapies may increase the risk of cardiovascular
disease.
 Poor diet. A diet that's high in fat, salt, sugar and cholesterol can contribute to the
development of heart disease.

15
 Heart Disease Prediction using Machine Learning Algorithms

 High blood pressure. Uncontrolled high blood pressure can result in hardening and
thickening of your arteries, narrowing the vessels through which blood flows.
 High blood cholesterol levels. High levels of cholesterol in your blood can increase
the risk of formation of plaques and atherosclerosis.
 Diabetes. Diabetes increases your risk of heart disease. Both conditions share similar
risk factors, such as obesity and high blood pressure.
 Obesity. Excess weight typically worsens other risk factors.
 Physical inactivity. Lack of exercise also is associated with many forms of heart
disease and some of its other risk factors, as well.
 Stress. Unrelieved stress may damage your arteries and worsen other risk factors for
heart disease.
 Poor hygiene. Not regularly washing your hands and not establishing other habits
that can help prevent viral or bacterial infections can put you at risk of heart
infections, especially if you already have an underlying heart condition. Poor dental
health also may contribute to heart disease.

1.4.6 Complications

Complications of heart disease include:

 Heart failure. One of the most common complications of heart disease, heart failure
occurs when your heart can't pump enough blood to meet your body's needs. Heart
failure can result from many forms of heart disease, including heart defects,
cardiovascular disease, valvular heart disease, heart infections or cardiomyopathy.
 Heart attack. A blood clot blocking the blood flow through a blood vessel that feeds
the heart causes a heart attack, possibly damaging or destroying a part of the heart
muscle. Atherosclerosis can cause a heart attack.
 Stroke. The risk factors that lead to cardiovascular disease also can lead to an
ischemic stroke, which happens when the arteries to your brain are narrowed or
blocked so that too little blood reaches your brain. A stroke is a medical emergency
— brain tissue begins to die within just a few minutes of a stroke.
 Aneurysm. A serious complication that can occur anywhere in your body, an
aneurysm is a bulge in the wall of your artery. If an aneurysm bursts, you may face
life-threatening internal bleeding.

16
 Heart Disease Prediction using Machine Learning Algorithms

 Peripheral artery disease. Atherosclerosis also can lead to peripheral artery disease.
When you develop peripheral artery disease, your extremities — usually your legs —
don't receive enough blood flow. This causes symptoms, most notably leg pain when
walking (claudication).
 Sudden cardiac arrest. Sudden cardiac arrest is the sudden, unexpected loss of heart
function, breathing and consciousness, often caused by an arrhythmia. Sudden cardiac
arrest is a medical emergency. If not treated immediately, it is fatal, resulting in
sudden cardiac death.

1.4.7 Prevention

Certain types of heart disease, such as heart defects, can't be prevented. However, you can
help prevent many other types of heart disease by making the same lifestyle changes that can
improve your heart disease, such as:

 Quit smoking
 Control other health conditions, such as high blood pressure, high cholesterol and
diabetes
 Exercise at least 30 minutes a day on most days of the week
 Eat a diet that's low in salt and saturated fat
 Maintain a healthy weight
 Reduce and manage stress
 Practice good hygiene

1.5 Some of the attributes we used for Heart Disease Prediction and their
correlation to CVD ( Cardiovascular Diseases )

Below we have explained some of the key attributes we have taken in to consideration in our
dataset for predicting whether the given data leads to conclude the presence of heart disease.

These key attributes are the very facts that has been used in determining a presence of heart
disease. Thus, here we shall be getting into deeper in checking how these factors relate to or
even cause Heart Diseases or CVD.

17
 Heart Disease Prediction using Machine Learning Algorithms

1.5.1 Age as a Cardiovascular Risk Factor

According to the most recent estimates from United States, cardiovascular disease (CVD) death
rates have declined but the disease burden still remains substantially high.[42] The risk of
developing CVD is largely (75–90%) explained by the presence or absence of traditional CVD
risk factors.[43] Age is a well known traditional risk factor, which is generally considered to
be non-modifiable for obvious reasons. In this review, we discuss the common use of an
individual’s age in prediction of CVD incidence using different risk scores, examine whether
age as a risk factor can be modified or not, discuss the methods used to evaluate long- and
short-term CVD risk, appropriate communication of an individual’s risk based on their age
group and CVD risk, and conclude by discussing the influence of age on cardiac and vascular
risk factors.

18
 Heart Disease Prediction using Machine Learning Algorithms

Assessment of CVD risk using Age as part of Risk Scores - With aging, there is an
incremental acquisition of several CVD risk factors in an individual’s lifespan. When these risk
factors are incorporated in a multivariable regression model, age still remains an independent
risk factor. There are several risk prediction scores currently available to assess an individual’s
risk of CVD, and all of them include ‘age’ as a predictor. Older age, as assessed by these risk
scores, is associated with greater risk of CVD. Although there are several risk scores available,
the Framingham Risk Score (FRS)[44] is one of the most-widely adopted screening tools in
United States and is recommended by National Heart Lung and Blood Institute to assess an
individual’s CVD risk. Other risk scores which are tested in Britain, Scotland, New Zealand or
China have not been formally tested in the United States. In addition to the traditional risk
factors (age, gender, smoking, total cholesterol, HDL-cholesterol and systolic blood pressure
which are part of FRS), risk scores developed in Britain and Scotland also incorporate family
history and social deprivation as risk factors, and these additional variables marginally improve
prediction of CVD risk over the FRS when applied to the British and the Scottish populations,
respectively. The Reynolds risk score also includes age as a component and is constructed
using a database of middle-aged American women and requires the additional measurements
of C-reactive protein and HbA1c (in diabetics).[45] Lastly, the risk prediction score reported
in prior European studies[46] and currently adopted by the Joint European societies[47] is
based on models which predict CVD death, and therefore underestimates the burden of CVD
by not including the non-fatal events. Note that although CVD death rates have declined in
some developed European countries (quite similar to the trend in the United States), the overall
CVD burden still remains high.[48]

Age is an Independent Risk Factor for Cardiovascular Disease - As discussed above, even
after adjusting for traditional risk factors in a multivariable CVD prediction model, age remains
a fundamental predictor of CVD risk.[48] However, when age and other risk factors are used
jointly to examine an individual’s future risk of CVD, it has been postulated that the
contribution of age in the multivariable models may be a reflection of the intensity and the
duration of exposure to other traditional CVD risk factors.[49] If this observation were true,
avoidance of these other risk factors should result in a reduction of CVD risk associated with
age per se. To examine this hypothesis prior studies from Framingham Heart Study have shown
that the absence of each of these traditional risk factors is associated with a reduction in the
risk of CVD even at an older age.[49] When the absence of multiple risk factors is factored

19
 Heart Disease Prediction using Machine Learning Algorithms

into an individual’s CVD risk assessment, the reduction in CVD risk is further augmented.
Similarly, using the Framingham cohort, investigators have observed that lower midlife blood
pressure and total cholesterol levels, absence of glucose intolerance, smoking abstinence,
higher education and female gender all predicted increased survival up to 85 years of age.[50]
Additionally at an older age, the contribution of age to CVD risk prediction declines, in part
because there is less time left for an individual to acquire other modifiable CVD risk factors.
Therefore, age at any given point influences the assessment of both the short- and long-term
CVD risk of an individual. The absence of these CVD risk factors not only prevents the
development of CVD but also decreases the risk of age-associated co-morbidities and
mortality.[51] In another prior study, after excluding individuals with cancer, cardiovascular
disease and diabetes before 50 years of age, investigators followed the Framingham cohort to
evaluate who was likely to reach 75 years of age. They concluded that smoking fewer cigarettes
per day, lower systolic blood pressure, and higher forced vital capacity were associated with
longevity in both sexes.[52] Moreover, these observations relating to presence and absence of
traditional risk factors have also been confirmed in a population-based study in the Japanese
cohort from the Honolulu Heart Program,[53] and the large scale, multiethnic and international
InterHeart Study.[54] The Inter Heart study investigators also tested this hypothesis in a case-
control fashion among all age groups and observed similar results for prevention of myocardial
infarction.[55] Therefore, it is now well established that life expectancy of an individual is
dependent on modification of traditional risk factors and age-associated risk of CVD can be
minimized by correcting or avoidance of these risk factors. Though, it is important to note that
risk factor modification is equally important for both young and older individuals and will
decrease their subsequent risk of CVD.[56]

Relative risk versus Absolute risk Assessment - Current CVD risk assessment using
Framingham risk score comprises of the traditional risk factors i.e. cholesterol (total and HDL),
blood pressure, history of smoking and age.[57] While assessing risk of CVD, it is important
that both short-term (10-year CVD risk) and long-term (>10 year) risk for CVD are evaluated,
and communicated appropriately to an individual.[58] At a younger age, an individual with
several CVD risk factors (i.e. smoker, increased cholesterol and high blood pressure) will have
a lower absolute short-term risk (compared to an older individual with similar CVD risk
factors), and the absolute risk increases as the person gets older. However, the relative risk
remains relatively invariant throughout a person’s lifespan provided other risk factors (except

20
 Heart Disease Prediction using Machine Learning Algorithms

age) do not change, and it may actually decrease over time. Similarly, an older individual with
several risk factors will have a higher short-term absolute risk (compared to a younger
individual with a similar risk factor profile) even though the relative risk may remain constant
through the lifespan, provided there is no change in risk factors.[59]

Communicating CVD Risk to Young and Old - Communicating either short- or long-term
CVD risk to a patient can be challenging and might over or under-estimate the importance of
risk factor reduction and therefore impact how a person would react by changing lifestyle for
future risk reduction. For example, communicating an overestimated relative risk to a young
individual might result in emotional or financial stress (may require them to take medications)
whereas communicating an under-estimated absolute risk may result in a lower level of
motivation on the part of an individual to work towards changing his/her lifestyle to reduce
CVD risk.21 Present guidelines from Adult Treatment Panel (ATP-III) for treatment of high
blood cholesterol appropriately incorporates both relative and absolute risk assessment aspects
(as discussed above) for an individual and provides flexibility for discussion by a treating
physician in primary prevention settings.[60] Prior investigators have cautioned treating
physicians to distance themselves from communicating the magnified relative risk of an
individual (compared to lower absolute risk) in order to achieve professionally desirable
goals.[61]

Influence of Age on Other Individual Risk Factors - It is intuitive that if age is an

independent risk factor for developing CVD, the lifetime risk of CVD for an individual would
continue to increase with age. However, the lifetime risk for CVD is lower at age 70 than at
age 50 years, for an individual whose lifestyle risk factors remains unchanged.[62] Similarly,
lifetime risk of coronary artery disease, stroke, hypertension and heart failure does not continue
to increase with age. One explanation for this observation is that there is shorter time period
left for older individuals to develop the disease and a greater hazard of death due to competing
causes. Other reasons are that those who live longer have inherent bias of lower burden of
cardiovascular risk factors which lowers their risk of developing an event, or a genetic makeup
with resistance to develop cardiovascular disease.[63] Framingham cohort enrolled individuals
at their midlife (30–62yrs) primarily but Inter Heart study included some young participants
(<40yrs) and both showed similar results that reduction or absence of risk factors is additive

21
 Heart Disease Prediction using Machine Learning Algorithms

and improves mortality. Consequently, it is important to note that screening for risk factors and
advice about modifications of risk factors should start at an early age.[64]

Influence of Individual Risk Factors on Age-associated CVD Risk - A sex-specific analysis

from Framingham cohort suggests about 11.9% ( men) to 40.3% (women) of age-associated
CVD risk may be attributable to the concomitant burden of other CVD risk factors.[65] These
estimates are based on comparing unadjusted regression coefficients for age with those
obtained after adjusting for other CVD risk factors in multivariable models (systolic blood
pressure, diabetes, total to high-density lipoprotein cholesterol ratio, history of smoking and
body mass index).[66]

1.5.2 Gender differences in coronary heart disease

Although CVD remains the leading killer of both women and men in the United States, there
are substantial sex/gender differences in the prevalence and burden of different CVDs, as
outlined. For both women and men, coronary heart disease (CHD) is the largest contributor to
CVD morbidity and mortality. The absolute numbers of women living with and dying of
CVD and stroke exceed those of men, as does the number of hospital discharges for heart
failure and stroke.[67] In 2007, women accounted for 60.6% of US stroke deaths.[68] In
contrast, more men are living with and dying of CHD than women and have more hospital
discharges for CVD and CHD. As shown in Figure 3, the prevalence of CHD is higher in
men within each age stratum until after 75 years of age, which may contribute to the
perception that heart disease is a man’s disease. Sex differences in CVD and CHD mortality
largely reflect sex differences in US demographics. Because female sex is associated with a
longer life expectancy than male sex, women constitute a larger proportion of the elderly
population in which the prevalence of CVD is greatest. Alarming statistics among younger
women 35 to 44 years of age show that CHD mortality rates have increased an average of
1.3% annually between 1997 and 2002, a statistically significant trend.[69]

22
 Heart Disease Prediction using Machine Learning Algorithms

Figure 3 Annual number of adults having diagnosed heart attack or fatal coronary heart disease by age and sex.

As illustrated in Figure 3 the absolute number of annual CVD deaths among the female sex
has exceeded that of the male sex since 1984. These data are often confused with CVD
mortality rates, which, when adjusted for differences in age distribution, reveal that the CVD
mortality rate is substantially higher in men than women. In 2007, the age-adjusted CVD
death rate in men was 300 per 100 000 compared with 212 per 100 000 women. The 2007
CVD mortality rate in women represents a 43% reduction from the rate in 1997. From 1980
to 2000, the age-adjusted death rate for CHD fell from 263 to 134 per 100 000 women;
during the same time period, the rate fell from 543 to 267 per 100 000 men.[70]

23
 Heart Disease Prediction using Machine Learning Algorithms

Figure 4 Trends in the total annual number of deaths caused by cardiovascular disease according to gender

The prevalence of CVD in women varies according to racial/ethnic minority status. The
prevalence of CVD among women ≥20 years of age is 47% among blacks, 34% among
whites, and 31% among Mexican Americans; the prevalence of CHD is 7.6%, 5.8%, and
5.6%, respectively.[70] Asian women ≥18 years of age have the lowest prevalence of CHD
(3.9%), according to the National Center for Health Statistics. The age-adjusted CHD death
rate is highest among black women (122 per 100 000 compared with 94 per 100 000 in white
women). The ominous trend for increasing rates of hypertension among black women is of
particular concern because the increased risk for both CHD and stroke compared with white
women could potentially widen the racial gap in CVD mortality. Dr Bernadine Healy first
introduced the concept of the Yentl syndrome in 1991, suggesting gender bias in the
management of CHD. There is ongoing debate as to whether women have a poorer prognosis
after a myocardial infarction (MI) than men, and why. Is any observed difference explained
by delay in women seeking care, healthcare provider delay in recognition and treatment,
underlying differences in pathophysiology, more comorbidities, or older ages at time of
presentation among women compared with men?[71]

24
 Heart Disease Prediction using Machine Learning Algorithms

Data over the past decade have shown that women have a higher 30-day mortality compared
with men, and it is now recognized that the gender differences are largely explained by
clinical differences at presentation. The higher mortality rate among women appears to be
limited primarily to ST-segment–elevation MI. It has also been suggested that higher death
rates may be restricted to younger women.[72] Although women with acute coronary
syndromes may have similar benefits from antiplatelet pharmacotherapy as men, they are
more likely to have bleeding problems, possibly as a result of excess dosing. Women
experience greater morbidity and mortality than men after coronary artery bypass grafting;
this disparity may reflect technical difficulties resulting from differences in body size, more
microvascular disease, and different risk factor profiles. More recently, it has been shown that
increasing use of off-pump coronary artery bypass grafting has narrowed the gender disparity
in outcomes. Early studies that examined gender differences in outcomes after MI and
revascularization may no longer be relevant owing to temporal trends in management and
risk factor profiles.[73] Recent data from the National Registry of Myocardial Infarction
showed that in-hospital mortality after an acute MI decreased more in women than in men
between 1994 and 2006; the absolute reduction was 3 times larger in women than in men <55
years of age (2.7% versus 0.9%). This narrowing of the mortality gap was explained largely
by greater improvements in risk factors in women than in men.[74] The classic risk factors
for CVD are the same in women and men, but there are gender differences in the prevalence
of risk factors. Although women and men overall have nearly equal percentages of
hypertension (1 in 3 adults), data from the National Health and Nutrition Examination Survey
(NHANES) showed that the prevalence of high blood pressure is greater in women >65 years
of age. The highest rate of hypertension is among black women, 44%, and is increasing. The
death rate caused by hypertension in 2007 was 37.0 per 100 000 for black women compared
with 14.3 per 100 000 for white women. Diabetes mellitus is more prevalent among women
than men ≥20 years of age (8.3% versus 7.2%).[75] Type II diabetes mellitus imparts a
greater risk of CHD in women than men and is not explained by differences in risk factors,
but rather by the more favorable survival rate of women (than men) without diabetes mellitus.
The prevalence of physician-diagnosed diabetes mellitus is highest among non-Hispanic
black (14.7%) and Mexican American (12.7%) women. On the basis of the NHANES data,
the age-adjusted prevalence of the metabolic syndrome is highest among Mexican American
women (40.6%), which is ≈22% higher than in Mexican American men. The prevalence of
total cholesterol ≥240 mg/dL in 2008 for those ≥20 years of age was 16.2% among women

25
 Heart Disease Prediction using Machine Learning Algorithms

and 13.5% among men. In contrast, the percent of women with high-density lipoprotein
cholesterol <40 mg/dL was 9.7% compared with 29.5% among men.[76]

Lifestyle risk factors also vary by gender, race, and ethnicity. Cigarette smoking has
decreased overall in the United States, but remains more common among men than women
(23.1% versus 18.1%). Non-Hispanic white women have a higher rate of smoking (20.7%)
than black women (18.8%) and Hispanic women (9.4%). Age-adjusted rates of physical
inactivity in 2009 were higher in women than men (34.5% versus 30.3%). Adverse trends in
levels of physical activity (> 12 times a month) reveal a decline from 1988 to 2006 from 57%
to 43% in men and from 49% to 43% in women.[77] The decreasing levels of physical
activity parallel the rising rates of overweight and obesity in the United States. Two thirds of
Americans are overweight or obese (72% of men and 64% of women) as defined by body
mass index. Among women, non-Hispanic blacks and Mexican Americans are more likely to
be obese than non-Hispanic whites (50% versus 45% versus 33%, respectively).[78] From
1999 to 2008, the increase in the prevalence of obesity was greater among men than
women.[79] Full adherence to 3 heart-healthy lifestyle behaviors (smoking abstinence,
physical activity, and fruit and vegetable intake) was nearly 50% higher among women than
men without CHD in a 2000 sample of the US population. Overall adherence was low
(<20%) for both women and men. These data suggest that population-wide approaches are
needed to reduce the burden of CVD in both genders.[80]

Closing the Gap in Preventive Care

Adherence to guidelines for the prevention of CVD is suboptimal for women and men. The
extent to which physician behaviors, patient behaviors, and environmental factors explain
nonadherence is not established.[81] The limited systematic evaluation of provider
performance in CVD preventive care makes it difficult to document gender differences in the
delivery of care. Etiologic explanations for any observed gender differences in adherence to
preventive recommendations are even more elusive. Most studies are conducted in select
settings, use a variety of quality indicators, and report limited data on confounding or effect-
modifying variables. Despite these research limitations, several themes consistently emerge
regarding barriers to optimal preventive care. A fundamental barrier to implementation of
prevention guidelines may be the guidelines themselves. Shaneyfelt et al evaluated the
guidelines process and found that longer guidelines included more standards than shorter

26
 Heart Disease Prediction using Machine Learning Algorithms

guidelines but were more often ignored in practice.[82] Evidence-based recommendations

were used more often than recommendations for practice not based on research evidence, and
controversial recommendations were followed less often than those that were
noncontroversial. A study of AHA/American College of Cardiology Guidelines showed that
adherence was higher when the recommendations were supported by randomized, controlled
clinical trials. Guidelines are more likely to be followed if they are easy to implement and
come from a highly respected source.[83] The AHA has published 3 women-specific
evidence-based guidelines between 2004 and 2011 for the prevention of CVD, but the extent
to which these guidelines changed physician behavior or affected any gender gap in risk
factor management is not known. The most recent AHA women’s guideline 2011 update
emphasized the importance of risk assessment to improve the quality of preventive care and
highlighted challenges of available risk assessment tools: short-term focus, relevance of
outcome measures (CVD versus CHD), and underestimation of risk in women. Further
research is needed to determine whether improved risk assessment is associated with
improved clinical outcomes.[84] Cabana et al evaluated 76 studies describing barriers to
adherence to clinical practice guidelines; lack of awareness, lack of familiarity, lack of
agreement, lack of self-efficacy, lack of outcome expectancy, and inertia of previous practice
were recurring thematic barriers for following guidelines. It was suggested that AHA
guidelines for the prevention of CVD in women are heterogeneous, and consequently there
are different barriers to implementation of individual recommendations.[85] In a national
AHA study of 500 randomly selected physicians, the most commonly cited barriers to
implementation of CVD prevention guidelines were time, insurance coverage, and the
patient. This study also revealed that physician assessment of CVD risk of the patient was the
primary driver of quality preventive care. Gender disparities in treatment were explained
largely by the provider’s lower perceived CVD risk in women, despite a similar calculated
risk compared with men. A subanalysis of this study suggested that solo practitioners and
older physicians should be targeted to help promote the use of the guidelines. In a program
designed to improve screening and management of CHD risk factors in women, internists and
obstetricians/gynecologists were queried about barriers to primary prevention; physician time
was perceived as a major barrier to the provision of preventive care.[86] The authors suggest
that the current structure and reimbursement system for health care must be addressed if the
gender gap in CVD preventive care is to be reduced. In a nationally representative sample of
women, the most frequently cited barriers to heart health were confusion in the media (49%),

27
 Heart Disease Prediction using Machine Learning Algorithms

the belief that health is determined by a higher power (44%), and caretaking responsibilities
(36%). Psychosocial factors may also contribute to nonadherence to preventive
recommendations in women. For example, depression and social isolation have been linked
to CVD risk and may be mediated by nonadherence to preventive recommendations, although
there is a lack of clinical trials to document that treatment of psychosocial risk improves
patient outcomes. The roles of body image and other psychological, social, and cultural
factors as mediators of nonadherence deserve further study. Systems approaches to CVD
prevention have the potential to improve outcomes and to reduce disparities. The Get With
the Guidelines Quality Improvement Program has shown improved adherence to secondary
prevention guidelines over time for both women and men, but the data are subject to selection
bias and secular trends.[87]

1.5.3 The association between blood pressure and mortality in patients with heart
failure.

Blood pressure is the force that pumps blood around the circulatory system. When blood flow
is restricted or blocked completely, the heart muscle is starved of oxygen. This leads to a heart
attack. During a heart attack, blood pressure can go up, down, or remain constant, depending
on how the body responds.[88]

Increase in blood pressure - Blood pressure might rise during a heart attack because
hormones, such as adrenaline, are released. These hormones are released when the "fight or
flight" response is triggered at times of intense stress or danger. This automatic response might
make the heart beat faster and stronger.[89]

Decrease in blood pressure - Blood pressure might drop if someone is having a heart attack
because the heart is too weak to maintain it, as the muscle might have been damaged. The
severe pain a person might feel during a heart attack could also trigger an automatic response,
which might lead to decreased blood pressure and fainting.[90]

Blood pressure and heart attacks - If high blood pressure is left untreated, it could increase
the risk of a heart attack. High blood pressure can be a measure of how hard the heart is having
to work to pump blood around the body via the arteries, which is why doctors monitor it. A
buildup of fat, cholesterol, and other substances within the arteries is called plaque. Over time,

28
 Heart Disease Prediction using Machine Learning Algorithms

plaque hardens, causing the arteries to narrow. This narrowing means it takes more pressure to
push the blood through the network of tubes. When plaque breaks away from the wall of an
artery, a blood clot is formed around the plaque. Heart attacks can happen because plaque or
blood clots cause the blood supply to the heart to be disrupted or blocked. High blood pressure
is not always a severe health problem, however. Even healthy people can experience raised
blood pressure from time to time due to exercise or stress.[91]

How is blood pressure measured?

1. Systolic blood pressure (SBP) is the pressure in the arteries, as the heart pumps blood
out to the body.
2. Diastolic blood pressure (DBP) is the pressure in the arteries between heart beats.

On blood pressure charts, the top number refers to the systolic pressure, while the number
underneath refers to the diastolic pressure.

The association between low blood pressure and prognosis in the general population has been
controversial, with some reports suggesting an increased mortality for patients with the lowest
blood pressures. Whereas many standard heart failure therapies decrease blood pressure, the
relationship between mortality and blood pressure in patients with heart failure has not been
previously evaluated. We used the Digitalis Investigation Group trial database to evaluate
retrospectively the relationship among systolic blood pressure (SBP), diastolic blood pressure
(DBP), and survival among 5747 patients with New York Heart Association class II or III heart
failure and left ventricular ejection fraction < or = 0.45. Cox proportional hazards models were
used to identify covariates predictive of long-term mortality.[92]

Figure 5 SBP and DBP wrt to Cardiovascular Disease Presence

29
 Heart Disease Prediction using Machine Learning Algorithms

RESULTS: The adjusted all-cause mortality rate during the entire study period for patients in
the lowest SBP group (< 100 mm Hg) was 50% and was significantly higher than that of the
reference group of patients with SBP of 130 to 139 mm Hg, which had a mortality rate of 32%
(hazard ratio 1.65, 95% CI 1.25-2.17, P < .001). The relationship between SBP and mortality
was significant (P < .001) and nonlinear (P = .009). The relationship between DBP and
mortality was significant (P < .001), with the highest mortality seen in patients with DBP < 60
mm Hg. In patients with systolic dysfunction (left ventricular ejection fraction < or = 0.45) and
New York Heart Association classes II and III symptoms, lower SBPs and DBPs were
associated with greater mortality.[93]

1.5.4 Chest Pain and its risk factor to Cardiac arrest

Angina is chest pain or discomfort caused when your heart muscle doesn't get enough oxygen-
rich blood. It may feel like pressure or squeezing in your chest. It is a symptom of an underlying
heart problem, usually coronary heart disease (CHD).There are many types of angina,
including microvascular angina, Prinzmetal's angina, stable angina, unstable angina and
variant angina. This usually happens because one or more of the coronary arteries is narrowed
or blocked, also called ischemia. Angina can also be a symptom of coronary microvascular
disease (MVD). This is heart disease that affects the heart’s smallest coronary arteries and is
more likely to affect women than men. Coronary MVD also is called cardiac syndrome X and
non-obstructive CHD. Learn more about angina in women.[94]

30
 Heart Disease Prediction using Machine Learning Algorithms

Figure 6 Pain Areas to be concerned about when having an Angina (Chest Pain)

Figure 7 Reversible and Progressive Angina

Types of Angina - Knowing the types of angina and how they differ is important.

 Stable Angina / Angina Pectoris

 Unstable Angina
 Variant (Prinzmetal) Angina
 Microvascular Angina

31
 Heart Disease Prediction using Machine Learning Algorithms

Diagnosis of Angina - All chest pain should be checked out by a healthcare provider. If you
have chest pain, your doctor will want to find out whether it's angina and if it is, whether the
angina is stable or unstable. If it's unstable, you may need emergency medical treatment to try
to prevent a heart attack.

Your doctor will most likely perform a physical exam, ask about your symptoms, and ask about
your risk factors for and your family history of heart disease and other cardiovascular
conditions.

1.5.5 Cholesterol and Heart Disease

Cholesterol helps your body build new cells, insulate nerves, and produce hormones. Normally,
the liver makes all the cholesterol the body needs. But cholesterol also enters your body from
food, such as animal-based foods like milk, eggs, and meat. Too much cholesterol in your body
is a risk factor for heart disease.[91]

How Does High Cholesterol Cause Heart Disease?

When there is too much cholesterol in your blood, it builds up in the walls of your arteries,
causing a process called atherosclerosis, a form of heart disease. The arteries become narrowed
and blood flow to the heart muscle is slowed down or blocked. The blood carries oxygen to the
heart, and if not enough blood and oxygen reach your heart, you may suffer chest pain. If the
blood supply to a portion of the heart is completely cut off by a blockage, the result is a heart
attack. There are two forms of cholesterol that many people are familiar with: Low-density
lipoprotein (LDL or "bad" cholesterol) and high-density lipoprotein (HDL or "good"
cholesterol.) These are the form in which cholesterol travels in the blood. LDL is the main
source of artery-clogging plaque. HDL actually works to clear cholesterol from the blood.
Triglycerides are another fat in our bloodstream. Research is now showing that high levels of
triglycerides may also be linked to heart disease.[95]

What Are the Symptoms of High Cholesterol?

High cholesterol itself does not cause any symptoms, so many people are unaware that their
cholesterol levels are too high. Therefore, it is important to find out what your cholesterol
numbers are. Lowering cholesterol levels that are too high lessens the risk for developing heart

32
 Heart Disease Prediction using Machine Learning Algorithms

disease and reduces the chance of a heart attack or dying of heart disease, even if you already
have it.[96]

Do I need Treatment For High Cholesterol?

Many health care providers recommend treating anyone with CVD with high-dose statin
therapy. This includes those with coronary heart disease and who have had a stroke. For those
who do not have CVD, treatment is determined by your individual risk for developing heart
disease. That risk can be estimated using calculators which factor your age, sex, medical
history, and other characteristics. If your risk is high (such as a 7.5 or 10 percent risk of
developing CVD over 10 years), your doctor may start you on treatment preventively. They
generally keep in mind your preferences towards taking medication in general. For those
people whose risk is unclear, a coronary artery calcium score, which is a screening test looking
for calcium (an indication of atherosclerosis) in the arteries, can help determine the need for
statins. For both those who have CVD and those who do not, when the decision is made to start
medication, the first choice is usually a statin.[97]

Other special groups who may need treatment:

 People with high triglyceride levels may benefit if they have other risk factors
 People with diabetes: are at high risk, and a ldl under 100 is recommended for most
 Older adults: a healthy, active older adult may benefit reduction you need, and prescribe
a medication accordingly.

1.5.6 Fasting Glucose Level or Fasting Blood Sugar and the Risk of Heart Diseases

Both low glucose level and impaired fasting glucose should be considered as predictors of risk
for stroke and coronary heart disease. The fasting glucose level associated with the lowest
cardiovascular risk may be in a narrow range.[98] Diabetes is a well-established risk factor for
cardiovascular disease (CVD) and all-cause mortality. Impaired fasting glucose (IFG), defined
by the American Diabetes Association as having a fasting plasma glucose level of 100–125
mg/dL (5.6–7.0 mmol/L) or a 2-h value on the oral glucose tolerance test of 140–199 mg/dL
(7.8–11.1 mmol/L) was associated with CVD risk in several studies.[99] The evidence is
inconsistent, however, and the clinical relevance of IFG as a predictor of CVD is still unclear.
In addition, the shape of the dose-response relationship between CVD risk and fasting glucose

33
 Heart Disease Prediction using Machine Learning Algorithms

level has not been well characterized across the full range of fasting blood glucose values. It is
unclear whether there is an optimum fasting glucose level associated with the lowest level of
CVD risk, or whether risk increases at very low fasting glucose levels.[95] Several studies have
shown J-shape or U-shape relationships between fasting glucose levels and mortality. The
Cancer Prevention Study (CPS) is a cohort study of >1.3 million adults designed to evaluate
major risk factors for chronic diseases and mortality. The large sample size of this cohort
facilitated detailed characterization of the dose-response relationship of fasting glucose level
with the incidence of clinical CVD end points. In a large cohort of men and women, we found
that fasting glucose level was associated with higher risk for major CVD outcomes, increasing
from a level of ∼90 mg/dL after controlling for other risk factors.[100] The dose-response
curves showed progressive increments in the HRs from this value at both higher and lower
levels; the increased risk was greatest for stroke. The patterns of association were similar in
men and women, but the associations were stronger in women. Substantial evidence supports
the biological plausibility of this finding. Experimental studies show that abnormal glucose
metabolism impairs normal endothelial function, accelerates atherosclerotic plaque formation,
and contributes to plaque rupture and thrombosis.[101] Epidemiological studies provide
complementary evidence. In the Rotterdam Study, among elderly participants with a fasting
blood glucose <110 mg/dL and without diabetes, those with higher blood glucose levels had
higher levels of arterial stiffness.

Figure 8 FBS analysis for Men

34
 Heart Disease Prediction using Machine Learning Algorithms

Figure 9 FBS analysis for Women

CATHAY study found that higher levels of glycemia (102–124 mg/dL) were associated with
arterial endothelial dysfunction and intima-media thickening. In a biomarker study in Italy, a
number of CVD biomarkers showed positive dose-response relationships with fasting glucose
across three strata: <100; 100–109; and 110–125 mg/dL. Our study adds to the increasing
evidence that IFG is an independent risk factor for incident CVD, including ischemic heart
disease and stroke. In addition, the effects of other CVD risk factors may be enhanced by
abnormal glucose metabolism.[102][103]

1.5.7 Electrocardiograph (ECG) Test for Heart Diseases

An electrocardiograph is the most common test for heart conditions. An electrocardiograph

machine records your heart's rhythm onto paper through sticky electrodes which are placed on
your chest, arms and legs. The recording will show if the heart muscle is damaged or short of
oxygen. Specialized ECG tests:

 An exercise tolerance test (ETT) involves two ECG scans, one when you are exercising
and one when you are resting. Some heart problems only appear when your heart needs
to work harder. This test helps to show how your heart copes under stress.[104]

35
 Heart Disease Prediction using Machine Learning Algorithms

 A cardiac Holter monitoring test is used to identify any heart rhythm problems. For this
test you wear a small, portable ECG machine for 24 or 48 hours and during this time
your heart rate and rhythm are recorded.[105]
 Event monitoring is used to record your heartbeat when you experience symptoms such
as dizziness, black outs, chest pain or palpitations. When you experience symptoms,
you will need to press a button to start the recording.[106]

1.5.8 Cardiac Complications in Thalassemia Major

Thalassemia major is characterized by chronic ineffective erythropoiesis and anemia as its

primary problems. These, in turn, produce physiologic adaptations in the cardiovascular system
as well as pathologic/iatrogenic processes such as iron overload, splenectomy, nutritional
deficiencies, chronic oxidative stress, and lung disease. This article discusses the
pathophysiology of thalassemia as it relates to the cardiovascular system, the mechanisms and
monitoring of iron cardiomyopathy, pulmonary hypertension, and vascular aging in
thalassemia patients.[107]

1. Chronic Anemia - Patients with chronic anemia increase their cardiac output to
maintain oxygen delivery, resulting in increased cardiac dimensions and heart rate.
Anemic patients have larger hearts on CXR, echo, and MRI measurements than patients
with normal hemoglobin levels, even without any other pathology. Thus, normative
data generated from non-anemic patients is inappropriate for patients with
hemoglobinopathies.[108] The larger cardiac dimensions, stroke volumes, and heart
rates carry metabolic cost; chronically anemic patients have higher resting oxygen
consumption and decreased reserves. Increased resting metabolism is also a source of
increased oxidative stress, independent of the free-radical effects of iron. Patients with
thalassemia have low or normal blood pressures, despite their increased cardiac output,
because they have lower vascular resistance. Lower tonic vascular tone partially
compensates for the increased chamber dimensions, but it leaves patients vulnerable to
the endothelial toxicity of iron overload as well as making them less tolerant and
responsive to the effects of afterload-reducing agents.
2. Splenectomy - Hypersplenism is relatively common in the thalassemia’s and may
necessitate spleen removal. Splenectomy may also be performed to lower blood

36
 Heart Disease Prediction using Machine Learning Algorithms

transfusion requirements. However, the spleen plays a critically important role in

removing hematologic debris from the cardiovascular system. Phosphatidylserine
positive platelets, platelet fragments, and red cell fragments are powerful pro-
coagulants. They also inhibit nitric oxide, stimulate vasoconstricting substances such
as endothelin and vasoconstricting prostaglandins, and produce endothelial
proliferation[109]. The spleen also removes brittle senescent red cells from the
circulation, suppressing intravascular hemolysis. Cell-free hemoglobin is a powerful
oxidant and scavenger of nitric oxide. As a result, splenectomy is a strong risk factor
for intravascular thrombosis and pulmonary hypertension.
3. Iron Overload - Patients with thalassemia develop iron overload through increased
iron absorption and trans fusional therapy. Iron is toxic to all the endocrine glands that
support the heart. Insulin-resistance and frank diabetes are relatively common.
Hyperglycemia and insulin resistance are powerful oxidative stressors to the heart,
worsening the effects of iron overload. Proper insulin sensitivity is also vital for
efficient cardiac energy utilization. Iron may also poison the thyroid and parathyroid
gland, impairing metabolism and calcium regulation respectively. Iron-mediated
adrenal insufficiency may also manifest itself during metabolic stress. Deficiencies of
growth hormone and the sex steroids impair cardiac function. Iron-mediated endocrine
toxicity must be excluded in TM patients with cardiac failure.[110]
4. Nutritional Deficiencies - The hemoglobinopathies are a hypermetabolic state and
inherently produce chronic oxidative stress. Broad-spectrum nutritional deficiencies are
prevalent and may reinforce disease toxicity. Fat-soluble vitamin depletion is common,
including vitamin A, D, E, and K, suggesting fat mal-absorption. The mechanisms and
consequences are unknown. Vitamin D deficiency is associated with increased cardiac
iron and decreased function, but causation has not been proven. Many trace metals are
decreased, including selenium, zinc, and copper. B-vitamin levels are also low,
particularly thiamine, riboflavin, and folate, most likely from consumption during
ineffective erythropoiesis. Severe thiamine deficiency can have neurological and
cardiac toxicity, whereas deficient riboflavin and folate may result in elevated
homocysteine and endothelial toxicity. Carnitine deficiency is also relatively common
in thalassemia and can impair cardiac function.[111]

37
 Heart Disease Prediction using Machine Learning Algorithms

Figure 10 Iron Cardiomyopathy - represents the pathophysiology of iron cardiomyopathy,

artificially divided into iron uptake, iron storage, and iron toxicity. The heart takes up physiologic
amounts of iron through transferrin receptors, but this process is tightly regulated and does not lead to
iron overload. When transferrin-binding capacity is exceeded, circulating low molecular weight non-
transferrin-bound iron (NTBI) species appear. NTBI is oxidatively active and can enter through
nonspecific, poor-regulated cation channels in the heart, leading to cardiac iron overload. Several
channel mechanisms have been proposed, including L-type voltage-dependent calcium channels, but
much more work is necessary to characterize cardiac iron-uptake processes.

38
 Heart Disease Prediction using Machine Learning Algorithms

Figure 11 Pulmonary Hypertension - demonstrates the complex pathophysiology of

pulmonary hypertension in thalassemia. Increased cardiac output and diastolic dysfunction
cause abnormal loading of the pulmonary artery. Lung disease can exacerbate night-time
hypoxia, a powerful stimulus for vasoconstriction. Iron, phosphatidylserine-expressing
hematologic debris, free hemoglobin, and other circulating angiotrophic factors cause
vasoconstriction and intimal proliferation.

39
 Heart Disease Prediction using Machine Learning Algorithms

CHAPTER: 2 RELATED WORKS ON HEART DISEASE

PREDICTION USING MACHINE LEARNING ALGORITHMS

2.1 Prediction system for heart disease using Naive Bayes and particle
swarm

This section provides the basic concepts of classifier as Naive Bayes and feature subset
selection method as PSO.

2.1.1 Particle swarm optimization (PSO)

PSO is an Evolutionary Computation technique is proposed by Kennedy et al. in 1995. PSO is

motivated by social behaviors such as bird flocking and fish schooling. In PSO population
swarm consists of “n” particles, and the position of each particle stands for the potential
solution in D-dimensional space. The particles change its condition based on three aspects: To
keep its inertia; T o change the condition according to its most optimist position; To change
the condition according to the swarm’s most optimist position[30]. In PSO, a population are
encoded as particles in the search space dimensionality D.PSO starts with the random
initialization of a population of particles. Based on the best experience of one particle
(pbest) and its neighboring particles (gbest), PSO searches for the optimal solution by
updating the velocity and the position of each particle; PSO is used as feature subset selection
method due to its advantages:
• Simple and easy to implement.
• Continuous optimization approach.

2.1.2 Naïve Bayes’ Classifier

Naive Bayes classifiers are a family of simple probabilistic classifiers based by using Bayes
theorem with s t r o n g (Naive) independence assumptions between the features. Naive Bayes
classifiers are highly scalable by requiring several parameters linear for the number of features
or predictors as variable in a learning problem.[112] It is the simplest and the fastest
probabilistic classifier especially for the training phase.

Feature selection - It is a process of removing the irrelevant and redundant features from
dataset based on evaluation criterion which is used to improve accuracy. There are two

40
 Heart Disease Prediction using Machine Learning Algorithms

approaches as individual evaluation and other one is subset evaluation. The process of
feature selection is classified into three broad classes. One is filter and another one is
wrapper and third one is embedded method based on how the feature selection is deployed
by supervised learning algorithm. In this paper, they proposed a model which uses Naive
Bayes as classifier and PSO as Feature subset selection measure for prediction of heart
disease.[113]
Proposed system - In this section, we propose a methodology to improve the performance of
Bayesian classifier for prediction of heart disease. Algorithm for our proposed model is shown
below:
Algorithm 1: Heart disease prediction by using Bayes classifier and PSO.
Input: Heart disease dataset.
Output: Classify patient dataset into heart disease or not (normal).

Step 1: Read the dataset.

Step 2: Apply particle swarm optimization for f e a t u r e selection.

Step 3: Remove the features with low value of PSO.

Step 4: Apply Naive Bayes classifier on relevant features. Step 5: Evaluate the
performance of NB+PSO model.
The above algorithm divided into two sections, section 1 (step 2 and step 3) performs
processing and feature subset selection. In section 2 (step 4 and step 5) Naive Bayes is
applied on relevant features data and evaluate the performance in terms of accuracy.[114]

Accuracy= (No. of objects correctly classified/Total no. of objects in test set)

Cross validation technique used to split into training and testing data.

2.2 Predictive Data Mining for Medical Diagnosis: Heart Disease Prediction

The successful application of data mining in highly visible fields like e-business, marketing
and retail has led to its application in other industries and sectors However, there is a lack o f
effective analysis tools to discover hidden relationships and trends in data. Number of

41
 Heart Disease Prediction using Machine Learning Algorithms

experiment has been conducted to compare the performance of predictive data mining
technique on the same dataset and the outcome reveals that Decision Tree outperforms and
sometime Bayesian classification is having similar accuracy as of decision tree but other
predictive methods like KNN, Neural Networks, Classification based on clustering are not
performing well. The second conclusion is that the accuracy of the Decision Tree and Bayesian
Classification further improves after applying genetic algorithm to reduce the actual data size
to get the optimal subset of attribute enough for heart disease prediction.[9][115]

2.2.1 Data Mining in the Heart Disease Prediction.

Three different supervised machine learning algorithms i.e. Naive Bayes, K-NN, Decision
List algorithm have been used for analyzing the dataset in. Tanagra tool is used to
classify the data and the data is evaluated using 10-fold cross validation and the results are
compared. Tanagra is a data mining suite build around graphical user interface algorithms.
Decision Tree is a popular classifier which is simple and easy to implement. It requires no
domain knowledge or parameter setting and can handle high dimensional data. The results
obtained from Decision Trees are easier to read and interpret. The drill through feature to
access detailed patients‟ profiles is only available in Decision Trees. Naïve Bayes is a
statistical classifier which assumes no dependency between attributes. It attempts to maximize
the posterior probability in determining the class. The advantage of using naive Bayes is that
one can work with the naive Bayes model without using any Bayesian methods. Naive Bayes
classifiers have works well in many complex real-world situations. The k-nearest neighbor’s
algorithm (k-NN) is a method for classifying objects based on closest training data in the
feature space. k-NN is a type of instance-based learning. The k-nearest neighbor algorithm
is amongst the simplest of all machine learning algorithms. But the accuracy of the k-NN
algorithm can be severely degraded by the presence of noisy or irrelevant features, or if the
feature scales are not consistent with their importance. The experiment is performed using
training data set consists of 3000 instances with 14 different attributes. The dataset is divided
into two parts that is 70% of the data are used for training and 30% are used for testing.

42
 Heart Disease Prediction using Machine Learning Algorithms

2.2.2 Data Mining and Artificial Neural network

Intelligent Heart Disease Prediction System (IHDPS) using data mining techniques, namely,
Decision Trees, Naïve Bayes and Neural Network. is implemented in using .NET platform.
IHDPS i s Web-based, user-friendly, scalable, reliable and expandable system. It can also
answer complex “what if” queries which traditional decision support systems cannot. Using
medical profiles such as age, sex, blood pressure and blood sugar it can predict the likelihood
of patients getting a heart disease. It enables significant knowledge, e.g. patterns, relationships
between medical factors related to heart disease. As a Data source a total of 909 records
with 15 medical attributes (factors) were obtained from the Cleveland Heart Disease database.
Figure 1 lists the attributes. The records were split equally into two datasets: training dataset
(455 records) and testing dataset (454 records). Initially, the data warehouse is pre- processed
in order to make it suitable for the mining process. Once the preprocessing gets over, the heart
disease warehouse is clustered with the aid of the K-means clustering algorithm, which will
extract the data appropriate to heart attack from the warehouse. Consequently, the frequent
patterns applicable to heart disease are mined with the aid of the MAFIA algorithm from the
data extracted. In addition, the patterns vital to heart attack prediction are selected on basis
of the computed significant weightage. The neural network is trained with the selected
significant patterns for the effective prediction of heart attack. Multi-layer Perceptron Neural
Network with Back-propagation is being used as the training algorithm. In feed-forward neural
networks the neurons of the first layer forward their output to the neurons of the second layer,
in a unidirectional fashion, which explains that the neurons are not received from the reverse
direction.[116][24]

43
 Heart Disease Prediction using Machine Learning Algorithms

Figure 12 The sample combinations of heart attack parameters for normal and risk level along with their values
and weightages

2.2.3 Data Mining and Genetic Algorithm

Genetic algorithm has been used in, to reduce the actual data size to get the optimal subset of
attributed sufficient for heart disease prediction. Classification is a supervised learning method
to extract models describing important data classes or to predict future trends. Three
classifiers e.g. Decision Tree, Naïve Bayes and Classification via clustering have been used to

44
 Heart Disease Prediction using Machine Learning Algorithms

diagnose the presence of heart disease in patients. Classification via clustering: Clustering
is the process of grouping similar elements. This technique may be used as a preprocessing
step before feeding the data to the classifying model. The attribute values need to be
normalized before clustering to avoid high value attributes dominating the low value attributes.
Further, classification is performed based on clustering. Experiments were conducted with
Weka 3.6.0 tool. Data set of 909 records with 13 attributes. All attributes are made categorical
and inconsistencies are resolved for simplicity. To enhance the prediction of classifiers,
genetic search is incorporated. Observations exhibit that the Decision Tree data mining
technique outperforms other two data mining techniques after incorporating feature subset
selection but with high model construction time. Naïve Bayes performs consistently before
and after reduction of attributes with the same model construction time.[117]

2.2.4 Association Rule Discovery

Association rules represent a promising technique to im prov e heart disease prediction.

Unfortunately, when association rules are applied on a medical data set, they produce an
extremely large number of rules. Most of such rules are medically irrelevant and the time
required to find them can be impractical. In, four constraints were proposed to reduce the
number of rules: item filtering, attribute grouping, maximum itemset size, and
antecedent/consequent rule filtering. When association rules are applied on a medical data set,
they produce an extremely large number of rules. Most of such rules are medically irrelevant
and the time required to find them can be impractical. A more important issue is that, in
general, association rules are mined on the entire data set without validation on an
independent sample. To solve these limitations, the author has introduced an algorithm that
uses search constraints to reduce the number of rules, searches for association rules on a
training set, and finally validates them on an independent test set. Instead of using only
Support and confidence, one more parameter i.e. lift have been used as the metrics to evaluate
the medical significance and reliability of association rules. Medical doctors use sensitivity
and specificity as two basic statistics to validate results. Sensitivity is defined as the
probability of correctly identifying sick patients, whereas specificity is defined as the
probability of correctly identifying healthy individuals. Lift was used together with
confidence to understand sensitivity and specificity. To find predictive association rules in a
medical data set the algorithm has three major steps. First, a medical data set with

45
 Heart Disease Prediction using Machine Learning Algorithms

categorical and numeric attributes is transformed into a transaction data set. Second, four
constraints mentioned above are incorporated into the search process to find predictive
association rules with medically relevant attribute combinations. Third, a train and test
approach are used to validate association rules.[118][119]

2.2.5 Issues and Challenges

Medical diagnosis is considered as a significant yet intricate task that needs to be carried out
precisely and efficiently. The automation of the same would be highly beneficial. Clinical
decisions are often made based on doctor’s intuition and experience rather than on the
knowledge rich data hidden in the database. This practice leads to unwanted biases, errors
and excessive medical costs which affects the quality of service provided to patients. Data
mining have the potential to generate a knowledge-rich environment which can help to
significantly improve the quality of clinical decisions.

46
 Heart Disease Prediction using Machine Learning Algorithms

CHAPTER 3: PRESENT
WORK

3.1 About Heart Disease

Heart Diseases affect a large population in today’s world, where the lifestyle is moved from
active to comfort-oriented. We live in era of fast foods. Which build up cholesterol, diabetes
and many more factors which in turn affects the heart in some way or the other. According to
the World Health Organization Cardiovascular Diseases (CVD) or Heart Diseases cause more
death than any other diseases globally[99]. The amount of data in medical sectors is quite large
and computerized as well. They are not utilized or put to any use. This data if studied and
analyzed could be put to good use like prediction of diseases or even prevent them. Diseases
such as cancer can be detected, and the stage can also be predicted by training dataset with
pictures of cancer cells. Similarly, heart disease can be predicted based on aspects like
cholesterol, diabetes, heart rate etc. The prediction of heart diseases is a challenge and very
risky. We observed that in some cases solutions of problems does not rely on a single method.
It varies from situation to situation. It is also a challenge as most of the data are sparse or
missing as they were not stored in the motive of analyzing. We therefore set out goal to finding
which method would be best for predicting the diseases using data of four different hospitals
from four different places. This is a comparative study on the efficiency of different data mining
techniques such as Logical Regression, Random forest, K-Nearest Neighbors, Decision Tree
in predicting heart diseases. The Data Mining techniques are analyzed, and the accuracy of
prediction is noted for each method used. The result showed that heart diseases can be predicted
with accuracy of above 90%. Cardiovascular diseases are the leading cause of death globally,
resulted in 17.9 million deaths (32.1%) in 2015, up from 12.3 million (25.8%) in 1990. It is
estimated that 90% of CVD is preventable. There are many risk factors for heart diseases that
we will take a closer look at. The main objective of this study is to build a model that
can predict the heart disease occurrence, based on a combination of features (risk factors)
describing the disease. Different machine learning techniques will be implemented and
compared upon standard performance metric such as accuracy. The dataset used for this study
was taken from UCI machine learning repository[120].

47
 Heart Disease Prediction using Machine Learning Algorithms

3.2 Motivation
The rate of heart diseases is increasing at an exponential rate. The busy lifestyle of people in
this era with all the fast food in the lunch break and getting back to sitting and working has
pushed as over the edge. Along with this people today have a lack of exercise and are less
active. For most of them recreation is just another movie in bed or anything technology
based. Physical activities have reduced drastically. These factors boosted the rate of heart
diseases to an unfortunately high percentage. In a developing country like ours the rate of
heart diseases has the same effect. The annual mortality rate per 100,000 people from
cardiovascular diseases in India has increased by 128.9% since 1990, an average of 5.6% a
year[121].

Prediction of heart diseases is a difficult and risky task. Since it is directly dependent on
people's’ health, accuracy is a major factor. If not predicted accurately it can be disastrous.
This research therefore focuses on the comparison of different data mining techniques to
predict it. It shows the comparative analysis of the different methods. Cross validation error is
used to compare the techniques. We choose Logical Regression, Random forest, K-Nearest
Neighbors, Decision Tree as they are the most widely used techniques in determining
diseases.

3.3 Proposal

In this project we have used 5 algorithms to find out the reasons of heart disease and create a
model to get the maximum accuracy possible. For this we have used UCI dataset[122] where
4 databases: Cleveland, Hungary, Switzerland, and the VA Long Beach are already stored
and the Creators are:

48
 Heart Disease Prediction using Machine Learning Algorithms

1. Hungarian Institute of Cardiology. Budapest: Andras Janosi, M.D.

2. University Hospital, Zurich, Switzerland: William Steinbrunn, M.D.
3. University Hospital, Basel, Switzerland: Matthias Pfisterer, M.D.
4. V.A. Medical Center, Long Beach and Cleveland Clinic Foundation: Robert Detrano,
M.D., Ph.D.
And the Donor is: David W. Aha (aha '@' ics.uci.edu) (714) 856-8779. From this dataset we
have used 5 widely used algorithms Logical Regression, Random forest, K-Nearest
Neighbors and Decision Tree to create the model with the maximum accuracy possible. We
have also explored precision score, recall score, F-score, false negative using confusion
matrix for every algorithm used.[123]

3.4 Dataset Structure & Description

3.4.1 Importing libraries

49
 Heart Disease Prediction using Machine Learning Algorithms

3.4.2 Load data

3.4.3 Check the type of the dataset

3.4.4 Check the Shape of the data

3.4.5 Check the top four columns of the dataset

50
 Heart Disease Prediction using Machine Learning Algorithms

3.4.6: Dataset description

The dataset used in this project contains 14 variables. The independent variable that needs to
be predicted, 'diagnosis', determines whether a person is healthy or suffer from heart disease.
Experiments with the Cleveland database have concentrated on endeavors to distinguish
disease presence (values 1, 2, 3, 4) from absence (value 0). There are several missing attribute
values, distinguished with symbol '?'. The header row is missing in this dataset, so the column
names have to be inserted manually.[124]
Features information:
 age - age in years
 sex - sex (1 = male; 0 = female)
 chest pain - chest pain type (1 = typical angina; 2 = atypical angina; 3 = non-
anginal pain; 4 = asymptomatic)
 blood pressure - resting blood pressure (in mm Hg on admission to the hospital)

51
 Heart Disease Prediction using Machine Learning Algorithms

 serum cholesterol - serum cholesterol in mg/dl

 fasting blood sugar - fasting blood sugar > 120 mg/dl (1 = true; 0 = false)
 electrocardiographic - resting electrocardiographic results (0 = normal; 1 = having
ST-T; 2 = hypertrophy)
 max heart rate - maximum heart rate achieved
 induced angina - exercise induced angina (1 = yes; 0 = no)
 ST depression - ST depression induced by exercise relative to rest
 slope - the slope of the peak exercise ST segment (1 = upsloping; 2 = flat; 3 = down
sloping)
 no of vessels - number of major vessels (0-3) colored by fluoroscopy
 thalassemia - 3 = normal; 6 = fixed defect; 7 = reversable defect
 diagnosis - the predicted attribute - diagnosis of heart disease (angiographic disease
status) (Value 0 = < 50% diameter narrowing; Value 1 = > 50% diameter
narrowing)

3.4.7 Types of features

Categorical features (Has two or more categories and each value in that feature can be
categorized by them): sex, chest pain
Ordinal features (Variable having relative ordering or sorting between the values): fasting
blood sugar, electrocardiographic, induced angina, slope, no of vessels, thalassemia,
diagnosis
Continuous features (Variable taking values between any two points or between the minimum
or maximum values in the feature column): age, blood pressure, serum cholesterol, max
heart rate, ST depression

52
 Heart Disease Prediction using Machine Learning Algorithms

3.4.8 Some Random data columns

3.4.9 Check for missing Data

53
 Heart Disease Prediction using Machine Learning Algorithms

No Data is missing, which is good.

3.4.10 Check the correlation with target data

3.5 Exploratory Data Analysis (EDA)

(1 is who have Heart Disease and 0 is who don’t have Heart Disease)

54
 Heart Disease Prediction using Machine Learning Algorithms

No. of Heart Disease patients is 165. No. of patients who don’t have a heart disease is 138.
[Which is a good balance of target data.]

Figure 13 Disease vs Non-Disease Analysis

3.5.1 Percentage of patient with or without heart problems in the given dataset

which is a good data distribution

55
 Heart Disease Prediction using Machine Learning Algorithms

3.5.2 Uniqueness of sex column –

Two sex types: 1 is male and 0 is female

3.5.3 Check the percentage and plot the graph

Figure 13 Sex analysis

56
 Heart Disease Prediction using Machine Learning Algorithms

3.5.4 Heart Disease Frequency for Ages

57
 Heart Disease Prediction using Machine Learning Algorithms

Figure 14 Heart Disease Frequency for ages

58
 Heart Disease Prediction using Machine Learning Algorithms

3.5.5 Heart Disease Frequency for sex

Where 1 is “male”, 0 is “female”.

Figure 15 Heart Disease Frequency for sex

59
 Heart Disease Prediction using Machine Learning Algorithms

3.5.6 Making the data column names easily recognizable

3.5.7 Checking out Male/Female Heart disease according to Fasting Blood Sugar

Fasting blood sugar test: A blood sample will be taken after an overnight fast. A fasting
blood sugar level less than 100 mg/dL (5.6 mmol/L) is normal. A fasting blood sugar level
from 100 to 125 mg/dL (5.6 to 6.9 mmol/L) is considered prediabetes. If it's 126 mg/dL (7
mmol/L) or higher on two separate tests, you have diabetes[125].

60
 Heart Disease Prediction using Machine Learning Algorithms

Figure 16 Heart Disease according to FBS

61
 Heart Disease Prediction using Machine Learning Algorithms

3.5.8 Analyzing the chest pain

There are four types of Angina(chest pain)[126].

[Value 1: typical angina[122], Value 2: atypical angina[127], Value 3: non-anginal pain[123],

Value 4: asymptomatic[128] ].

Let’s check how many types of chest pain (Angina) is present in our dataset

4 types are present, 0, 1, 2, 3

Let’s plot chest pain types against target

62
 Heart Disease Prediction using Machine Learning Algorithms

Figure 17 Chest Pain Analysis

3.5.9 Analyzing the resting blood pressure

Resting blood pressure in mm Hg on admission to the hospital [129][124].

63
 Heart Disease Prediction using Machine Learning Algorithms

Let’s check out the unique resting blood pressures in our dataset

Let’s plot the resting blood pressure of our dataset against target

64
 Heart Disease Prediction using Machine Learning Algorithms

Figure 18 Resting Blood Pressure Analysis

65
 Heart Disease Prediction using Machine Learning Algorithms

3.5.10 Analyzing the resting electrocardiographic measurement

0 = normal, 1 = having ST-T wave abnormality, 2 = showing probable or definite left

ventricular hypertrophy by Estes' criteria[130][131][132].

Let’s check uniqueness of our dataset

There are 3 types of resting ECG values are present: 0, 1, 2.

Let’s then plot the ECG values against target column

People with resting ECG value: 1 and 0 are much likely to have a heart disease than
with the value 2 of resting ECG.

66
 Heart Disease Prediction using Machine Learning Algorithms

Figure 19 ECG analysis

3.5.11 Analyzing Exercise Induced angina

1 means yes, and 0 means no.

67
 Heart Disease Prediction using Machine Learning Algorithms

Figure 20 Exercised Induced Angina Analysis

3.5.12 Slope of the peak exercise ST segment

The treadmill electrocardiogram (ECG) stress test is widely used to screen for obstructive
coronary artery disease (CAD). The presence of ST segment changes, either depression or
elevation, on the ECG during the treadmill test often suggests presence of CAD and warrants
further management. We herein present three cases, with evidence of ischemia on the

68
 Heart Disease Prediction using Machine Learning Algorithms

treadmill ECG stress test. In addition, we discuss the use of the treadmill ECG stress test,
including its indications, contraindications, reasons for termination and interpretation of the
ST-segment changes, heart rate, as well as blood pressure responses to exercise[133].

Value 1: upsloping, Value 2: flat, Value 3: down sloping.

Slope '2' causes heart pain much more than Slope '0' and '1'.

69
 Heart Disease Prediction using Machine Learning Algorithms

Figure 21 ST Slope Analysis

70
 Heart Disease Prediction using Machine Learning Algorithms

3.5.13 Analyzing no. of major vessels colored by fluoroscopy

Figure 22 Analyzing no. of major vessels colored by fluoroscopy

71
 Heart Disease Prediction using Machine Learning Algorithms

3.5.13.1 Comparing with target:

Figure 23 Comparing with targets

3.5.14 Analyzing thalassemia

Four alpha-globin and two beta-globin protein chains make up hemoglobin. The two main
types of thalassemia are alpha and beta.

72
 Heart Disease Prediction using Machine Learning Algorithms

Alpha thalassemia - In alpha thalassemia, the hemoglobin does not produce enough alpha
protein. To make alpha-globin protein chains we need four genes, two on each chromosome
16. We get two from each parent. If one or more of these genes is missing, alpha thalassemia
will result. The severity of thalassemia depends on how many genes are faulty, or mutated.

 One faulty gene: The patient has no symptoms. A healthy person who has a child
with symptoms of thalassemia is a carrier. This type is known as alpha thalassemia
minima.

 Two faulty genes: The patient has mild anemia. It is known as alpha thalassemia
minor.

 Three faulty genes: The patient has hemoglobin H disease, a type of chronic anemia.
They will need regular blood transfusions throughout their life.

 Four faulty genes: Alpha thalassemia major is the most severe form of alpha
thalassemia. It is known to cause hydrops fetalis, a serious condition in which fluid
accumulates in parts of the fetus' body. A fetus with four mutated genes cannot
produce normal hemoglobin and is unlikely to survive, even with blood transfusions.
Alpha thalassemia is common in southern China, Southeast Asia, India, the Middle
East, and Africa.

Beta Thalassemia - We need two globin genes to make beta-globin chains, one from each
parent. If one or both genes are faulty, beta thalassemia will occur.
Severity depends on how many genes are mutated.

 One faulty gene: This is called beta thalassemia minor.

 Two faulty genes: There may be moderate or severe symptoms. This is known as
thalassemia major. It used to be called Colley's anemia.

Beta thalassemia is more common among people of Mediterranean ancestry. Prevalence is

higher in North Africa, West Asia, and the Maldives Islands. So, we’ll mainly work with

73
 Heart Disease Prediction using Machine Learning Algorithms

Alpha Thalassemia, And value 0 is one faulty gene, Value 1 is Two faulty genes, Value 2 is
Three faulty genes, Value 3 is Four faulty genes.

Thalassemia distribution

Figure 24 Thalassemia distribution

74
 Heart Disease Prediction using Machine Learning Algorithms

Against target:

Figure 25 Thalassemia against target

75
 Heart Disease Prediction using Machine Learning Algorithms

3.5.15 Thalassemia vs cholesterol

Figure 26 Thalassemia vs cholesterol

76
 Heart Disease Prediction using Machine Learning Algorithms

Thalassemia vs resting blood pressure

Figure 27 Thalassemia vs resting blood pressure

77
 Heart Disease Prediction using Machine Learning Algorithms

3.6 Correlation Matrix

Correlation analysis is a method of statistical evaluation used to study the strength of a
relationship between two, numerically measured, continuous variables (e.g. height and
weight)

Store numeric variables in cname variable

Let’s plot the Correlation plot

Figure 28 Correlation Matrix

78
 Heart Disease Prediction using Machine Learning Algorithms

Correlation Analysis:

There is no single feature that has a very high correlation with our target value. Also, some of
the features have a negative correlation with the target value and some have positive.

79
 Heart Disease Prediction using Machine Learning Algorithms

3.7 Data Preparation

Total Among 303 data’s randomly 242 are chosen for Training and 61 are chosen for Testing.

3.8 Modelling and predicting with Machine Learning

The main goal of the entire project is to predict heart disease occurrence with the highest
accuracy. In order to achieve this, we will test several classification algorithms. This section
includes all results obtained from the study and introduces the best performer according to
accuracy metric. We have chosen several algorithms typical for solving supervised learning
problems throughout classification methods.

Modelling and predicting with Machine Learning

The main goal of the entire project is to predict heart disease occurrence with the highest
accuracy. In order to achieve this, we will test several classification algorithms. This section
includes all results obtained from the study and introduces the best performer according to
accuracy metric. We have chosen several algorithms typical for solving supervised learning
problems throughout classification methods.

80
 Heart Disease Prediction using Machine Learning Algorithms

First of all, let's equip ourselves with a handy tool that benefits from the cohesion of SciKit
Learn library and formulate a general function for training our models. The reason for
displaying accuracy on both, train and test sets, is to allow us to evaluate whether the model
overfits or underfits the data (so-called bias/variance tradeoff).

3.8.1 Logistic Regression

Logistic regression is a classification algorithm used to assign observations to a discrete set of
classes. Unlike linear regression which outputs continuous number values, logistic regression
transforms its output using the logistic sigmoid function to return a probability value which
can then be mapped to two or more discrete classes.

Types of logical regression:

 Binary (Pass/Fail)
 Multi (Cats, Dogs, Sheep)

Sigmoid function

S(z)=1/1+e^−z

81
 Heart Disease Prediction using Machine Learning Algorithms

Figure 29 sigmoid function

Decision Boundary

p≥0.5, class=1 p<0.5, class=0

Figure 30 Decision Boundary

Cost Function

82
 Heart Disease Prediction using Machine Learning Algorithms

Vectorized cost function

For Multiclass - Instead of y=0,1 we will expand our definition so that y=0,1...n. Basically
we re-run binary classification multiple times, once for each class.

Procedure -

1. Divide the problem into n+1 binary classification problem (+1 because the index starts at
0).
2. For each class…
3. Predict the probability the observations are in that single class.
4. prediction = <math>max (probability of the classes)

Accuracy score of Logistic Regression is: 85.25%

Confusion Matrix

83
 Heart Disease Prediction using Machine Learning Algorithms

Figure 31 logistic regression confusion matrix

3.8.1.2 precision score

Recall

84
 Heart Disease Prediction using Machine Learning Algorithms

F-Score

false negative

3.8.2 Random Forest

Random Forest is a supervised learning algorithm. Random forest can be used for both
classification and regression problems, by using random forest regressor we can use random
forest on regression problems. But we have used random forest on classification in this project
so we will only consider the classification part.

Random Forest pseudocode

1. Randomly select “k” features from total “m” features.

Where k << m

2. Among the “k” features, calculate the node “d” using the best split point.
3. Split the node into daughter nodes using the best split.
4. Repeat 1 to 3 steps until “l” number of nodes has been reached.

85
 Heart Disease Prediction using Machine Learning Algorithms

5. Build forest by repeating steps 1 to 4 for “n” number times to create “n” number of
trees.

Random forest prediction pseudocode

1. Takes the test features and use the rules of each randomly created decision tree to
predict the outcome and stores the predicted outcome (target)
2. Calculate the votes for each predicted target.
3. Consider the high voted predicted target as the final prediction from the random
forest algorithm.

Pruning the depth of the trees

Accuracy score of Random Forest is 86.9%

86
 Heart Disease Prediction using Machine Learning Algorithms

Confusion Matrix

Figure 32 Random forest confusion matrix

Precision score

Recall

F-Score

87
 Heart Disease Prediction using Machine Learning Algorithms

False negative rate

3.8.3 Naïve Bayes

Bayes’ Theorem is stated as:

P(h|d) = (P(d|h) * P(h)) / P(d)

 P(h|d) is the probability of hypothesis h given the data d. This is called the posterior
probability.
 P(d|h) is the probability of data d given that the hypothesis h was true.
 P(h) is the probability of hypothesis h being true (regardless of the data). This is called
the prior probability of h.
 P(d) is the probability of the data (regardless of the hypothesis).

we are interested in calculating the posterior probability of P(h|d) from the prior probability
p(h) with P(D) and P(d|h). After calculating the posterior probability for a number of different
hypotheses, we will select the hypothesis with the highest probability. This is the maximum
probable hypothesis and may formally be called the (MAP) hypothesis.
This can be written as:

MAP(h) = max(P(h|d))

or

MAP(h) = max((P(d|h) * P(h)) / P(d))

or

88
 Heart Disease Prediction using Machine Learning Algorithms

MAP(h) = max(P(d|h) * P(h))

The P(d) is a normalizing term which allows us to calculate the probability. We can drop it
when we are interested in the most probable hypothesis as it is constant and only used to
normalize. Back to classification, if we have an even number of instances in each class in our
training data, then the probability of each class (e.g. P(h)) will be equal. Again, this would be
a constant term in our equation, and we could drop it so that we end up with:

MAP(h) = max(P(d|h))

Naive Bayes is a classification algorithm for binary (two-class) and multi-class classification
problems. The technique is easiest to understand when described using binary or categorical
input values. It is called naive Bayes or idiot Bayes because the calculation of the
probabilities for each hypothesis are simplified to make their calculation tractable. Rather
than attempting to calculate the values of each attribute value P (d1, d2, d3|h), they are
assumed to be conditionally independent given the target value and calculated as P(d1|h) *
P(d2|H) and so on. This is a very strong assumption that is most unlikely in real data, i.e. that
the attributes do not interact. Nevertheless, the approach performs surprisingly well on data
where this assumption does not hold.

MAP(h) = max(P(d|h) * P(h))

Gaussian Naïve Bayes

mean(x) = 1/n * sum(x)

Where n is the number of instances and x are the values for an input variable in your training
data. We can calculate the standard deviation using the following equation:

standard deviation(x) = sqrt (1/n * sum(xi-mean(x)^2))

This is the square root of the average squared difference of each value of x from the mean
value of x, where n is the number of instances, sqrt() is the square root function, sum() is the
sum function, xi is a specific value of the x variable for the i’th instance and mean(x) is
described above, and ^2 is the square. Gaussian PDF with a new input for the variable, and in

89
 Heart Disease Prediction using Machine Learning Algorithms

return the Gaussian PDF will provide an estimate of the probability of that new input value
for that class.

pdf (x, mean, sd) = (1 / (sqrt (2 * PI) * sd)) * exp (-((x-mean^2)/(2*sd^2)))

Where pdf(x) is the Gaussian Probability Density Function (PDF), sqrt () is the square root,
mean and sd are the mean and standard deviation calculated above, Pi is the numerical
constant, exp () is the numerical constant e or Euler’s number raised to power and x is the
input value for the input variable.

Confusion matrix

Figure 33 Naive Bayes confusion matrix

Precision score

90
 Heart Disease Prediction using Machine Learning Algorithms

Recall

F-Score

False negative

3.8.4 K-Nearest Neighbor

We can implement a KNN model by following the below steps:

1. Load the data

2. Initialize the value of k
3. For getting the predicted class, iterate from 1 to total number of training data points

 Calculate the distance between test data and each row of training data. Here
we will use Euclidean distance as our distance metric since it’s the most
popular method. The other metrics that can be used are Chebyshev, cosine,
etc.
 Sort the calculated distances in ascending order based on distance values
 Get top k rows from the sorted array
 Get the most frequent class of these rows
 Return the predicted class

91
 Heart Disease Prediction using Machine Learning Algorithms

Using different inputs –

It turns out that the value of n = 8 is optimal

92
 Heart Disease Prediction using Machine Learning Algorithms

Confusion matrix

Figure 34 KNN confusion matrix

Precision score

Recall

F-Score

False negative

93
 Heart Disease Prediction using Machine Learning Algorithms

3.8.5 Decision Tree

Pseudocode

1. Place the best attribute of the dataset at the root of the tree.
2. Split the training set into subsets. Subsets should be made in such a way that each
subset contains data with the same value for an attribute.
3. Repeat step 1 and step 2 on each subset until you find leaf nodes in all the branches of
the tree.

Assumptions while creating Decision Tree

 At the beginning, the whole training set is considered as the root.

 Feature values are preferred to be categorical. If the values are continuous then they
are discretized prior to building the model.
 Records are distributed recursively on the basis of attribute values.
 Order to placing attributes as root or internal node of the tree is done by using some
statistical approach.

The popular attribute selection measures

 Information gain
 Gini index

94
 Heart Disease Prediction using Machine Learning Algorithms

Attribute selection method - A dataset consists of “n” attributes then deciding which attribute
to place at the root or at different levels of the tree as internal nodes is a complicated step. By
just randomly selecting any node to be the root can’t solve the issue. If we follow a random
approach, it may give us bad results with low accuracy. For solving this attribute selection
problem, researchers worked and devised some solutions. They suggested using
some criterion like information gain, Gini index, etc. These criterions will calculate values
for every attribute. The values are sorted, and attributes are placed in the tree by following the
order i.e., the attribute with a high value (in case of information gain) is placed at the root.
While using information Gain as a criterion, we assume attributes to be categorical, and for
Gini index, attributes are assumed to be continuous.[132]

Gini Index - Gini Index is a metric to measure how often a randomly chosen element would
be incorrectly identified. It means an attribute with lower Gini index should be preferred.

The accuracy on the training set is 100%, while the test set accuracy is much worse. This is
an indicative that the tree is overfitting and not generalizing well to new data. Therefore, we
need to apply pre-pruning to the tree. We set max depth=3, limiting the depth of the tree
decreases overfitting. This leads to a lower accuracy on the training set, but an improvement
on the test set.

95
 Heart Disease Prediction using Machine Learning Algorithms

Figure 35 Decision tree visualization

96
 Heart Disease Prediction using Machine Learning Algorithms

Confusion matrix

Figure 36 decision tree confusion matrix

Precision score

Recall

F-Score

False negative rate

97
 Heart Disease Prediction using Machine Learning Algorithms

3.9 Result

Figure 37 Accuracy Scores of all Algorithms

98
 Heart Disease Prediction using Machine Learning Algorithms

3.10 System Requirements

1. Intel® Core™ i5 processor 8250U at 1.60 GHz or 1.80 GHz, 8 GB of DRAM.

2. Disk space 2TB.
3. Operating System: 64-bit Windows 10 Pro.

Recommended System Requirements

 Processors: Intel® Core™ i5 processor 4300M at 2.60 GHz or 2.59 GHz (1 socket, 2
cores, 2 threads per core), 8 GB of DRAM.
 Operating systems: Windows® 10, macOS*, and Linux*

Minimum System Requirements

 Processors: Intel Atom® processor or Intel® Core™ i3 processor

 Disk space: 1 GB
 Operating systems: Windows* 7 or later, macOS, and Linux
 Python* versions: 2.7.X, 3.6.X
 Included development tools: conda*, conda-env, Jupyter Notebook* (IPython)
 Compatible tools: Microsoft Visual Studio*, PyCharm*Included Python packages -
NumPy, SciPy, scikit-learn*, pandas, Matplotlib, Numba*, Intel® Threading Building
Blocks, pyDAAL, Jupyter, mpi4py, PIP*, and others.

Software

 PIP and NumPy: Ubuntu*, Python 3.6.2, NumPy 1.13.1, scikit-learn 0.18.2
 Windows: Python 3.6.2, PIP and NumPy 1.13.1, scikit-learn 0.18.2
 Intel® Distribution for Python* 2018


Modifications

 Scikit-learn: Conda*-installed NumPy with Intel® Math Kernel Library (Intel® MKL)
on Windows (PIP-installed SciPy on Windows contains Intel MKL dependency).

99
 Heart Disease Prediction using Machine Learning Algorithms

CHAPTER: 4 CONCLUSION

4.1 Inference

The overall objective of our project is to predict accurately with less number of tests and
attributes the presence of heart disease. In this project, fourteen attributes are considered
which form the primary basis for tests and give accurate results more or less. Many more
input attributes can be taken but our goal is to predict with less number of attributes and
faster efficiency to predict the risk of having heart disease at a particular age span. Five data
mining classification techniques were applied namely K-Nearest Neighbor, Naive Bayes,
Decision Tree, Random Forest & Logistic Regression. It is shown that Random Forest has
better accuracy than the other techniques.

This is the most effective model to predict patients with heart disease. This project
could answer complex queries, each with its own strength with respect to ease of model
interpretation, access to detailed information and accuracy.

This project can be further enhanced and expanded. For example, it can incorporate
other medical attributes besides the 14 attributes we used. It can also incorporate other data
mining techniques, e.g., Time Series, Clustering and Association Rules. Continuous data can
also be used instead of just categorical data. Another area is to use Text Mining to mine the
vast amount of unstructured data available.

This project is presented using data mining techniques. From logistic regression,
KNN, Naive Bayes, Decision Tree, Random forest are used to develop the system. Random
Forest proves the better results and assists the domain experts and even the person related to
the medical field to plan for a better and early diagnosis for the patient. This system performs
realistically well even without retraining

100
 Heart Disease Prediction using Machine Learning Algorithms

ACTIVITY DIAGRAM

101
 Heart Disease Prediction using Machine Learning Algorithms

4.2 Drawbacks

The Algorithms used in our project does not give a 100% accuracy, so the prediction is not
100% feasible. Clinical diagnosis and diagnosis using our project may differ slightly because
the prediction is not 100% accurate. Medical diagnosis is considered as a significant yet
intricate task that needs to be carried out precisely and efficiently. The automation of the
same would be highly beneficial. Clinical decisions are often made based on doctor’s
intuition and experience rather than on the knowledge rich data collected from the dataset.

4.3 Future Scope

We are planning to introduce an efficient disease prediction system to predict the heart
disease with better accuracy using Support Vector Machine (SVM). Our project aims to
provide a web platform to predict the occurrences of disease based on various symptoms. The
user can select various symptoms and can find the diseases with their probabilistic figures.
Our project can be improved by implementing medicine suggestion to the patient along with
the results. We can implement a feedback from the experienced doctors who can give their
views and opinions about certain medicines /practices done by the doctor on the patient. We
can implement a live chat option where the patient can chat with a doctor available regarding
medication for the respective result for their symptoms. Our project could be used as a
training tool for Nurses and Doctors who are freshly introduced in the field related to heart
diseases. The patient can have a choice in choosing the medicines he/she should take in order
to have a healthier life. Moreover, if implemented on a large scale it can be used in medical
facilities like hospital, clinics where a patient wouldn’t have to wait in long queues for
treatment if he is feeling symptoms related to heart disease.

102
 Heart Disease Prediction using Machine Learning Algorithms

Sequence Diagram for Future Scope

103
 Heart Disease Prediction using Machine Learning Algorithms

REFERENCES

[1] M. K. Awang and F. Siraj, “Utilization of an artificial neural network in the prediction
of heart disease,” Int. J. Bio-Science Bio-Technology, vol. 5, no. 4, pp. 159–165, 2013.

[2] P. Selvakumar and S. P. Rajagopalan, “A survey on neural network models for heart
disease prediction,” J. Theor. Appl. Inf. Technol., vol. 67, no. 2, pp. 485–497, 2014.

[3] A. Methaila, P. Kansal, H. Arya, and P. Kumar, “Early Heart Disease Prediction Using
Data Mining Techniques,” vol. 6956, no. October, pp. 53–59, 2014.

[4] I. S. F. Dessai, “Intelligent Heart Disease Prediction System Using Probabilistic

Neural Network,” no. 5, pp. 38–44, 2013.

[5] T. Karayilan and Ö. Kiliç, “Prediction of Heart disease using neural network,” in 2nd
International Conference on Computer Science and Engineering, UBMK 2017, 2017,
pp. 719–723.

[6] M. Mardiyono, R. Suryanita, and A. Adnan, “Intelligent Monitoring System on

Prediction of Building Damage Index using Neural-Network,” TELKOMNIKA
(Telecommunication Comput. Electron. Control., vol. 10, no. 1, p. 155, 2015.

[7] N. Guru, A. Dahiya, and N. Rajpal, “Decision support system for heart disease
diagnosis using Neural Network,” 2007.

[8] J. S. Singh Navdeep, “No Title,” Intell. Hear. Dis. Predict. Syst. using CANFIS Genet.
Algorithm, Int. J. Biol. Med. Sci., no. International Journal of Advance Research, Ideas
and Innovations in Technology© 2018, www.IJARIIT.comAll Rights, p. Reserved
Page | 987, 2008.

[9] R. G. S. Rajkumar Asha, “No Title,” Diagnosis Hear. Dis. Using Datamining
Algorithm, vol. Issue 10 V, no. Global Journal of Computer Science and Technology,
p. Page 38, 2010.

[10] S. Radhimeenakshi and G. M. Nasira, “Prediction of Heart Disease using Neural

Network with Back Propagation,” Int. J. Data Min. Tech. Appl., vol. 4, no. 1, pp. 19–
22, 2016.

[11] “Prediction of Heart Disease using Artificial Neural Network,” VFAST Trans. Softw.
Eng., pp. 102–112, 2019.

104
 Heart Disease Prediction using Machine Learning Algorithms

[12] Z. M. A. Mehta Rupa G. ,Rana Dipti P., “No Title,” A Nov. Fuzzy Based Classif. Data
Min. using Fuzzy Discret., no. World Congress on Computer Science and Information,
2009.

[13] A. A. Shinde, S. N. Kale, R. M. Samant, A. S. Naik, and S. A. Ghorpade, “Heart

Disease Prediction System using Multilayered Feed Forward Neural Network and
Back Propagation Neural Network,” Int. J. Comput. Appl., vol. 166, no. 7, pp. 975–
8887, 2017.

[14] N. Al-Milli, “Backpropogation neural network for prediction of heart disease,” J.

Theor. Appl. Inf. Technol., vol. 56, no. 1, pp. 131–135, 2013.

[15] G. S. Kumari Milan, “No Title,” Comp. Study Data Min. Classif. Methods Cardiovasc.
Dis. Predict., vol. Vol. 2, no. IJCST Vol. 2, Iss ue2, June 2011, 2011.

[16] C. Ordonez, “Improving Heart Disease Prediction using Constrained Association

Rules,” Tech. Semin. Present. Univ. Tokyo, 2004.

[17] M. C. and P. M. Franck Le Duff, CristianMunteanb, “Predicting Survival Causes After

Out of Hospital Cardiac Arrest using Data Mining Method,” Stud. Health Technol.
Inform., vol. Vol. 107, no. 2, p. No. 2, pp. 1256–1259, 2004.

[18] W. J. F. and G. Piatetsky-Shapiro, “Knowledge Discovery in Databases: An

Overview,” AI Mag., vol. Vol. 13, N, no. 3, pp. 57–70, 1996.

[19] K. Y. N. and K. H. R. Heon Gyu Lee, “Mining Bio Signal Data: Coronary Artery
Disease Diagnosis using Linear and Nonlinear Features of HRV,” Proc. Int. Conf.
Emerg. Technol. Knowl. Discov. Data Min., p. pp. 56–66, 2007.

[20] B. J. L. and K. H. R. Kiyong Noh, HeonGyu Lee, Ho-Sun Shon, “Associative

Classification Approach for Diagnosing Cardiovascular Disease,” Intell. Comput.
Signal Process. Pattern Recognit., vol. 345, pp. 721–727, 2006.

[21] L. P. and R. Subramanian, “Intelligent Heart Disease Prediction System using

CANFIS and Genetic Algorithm,” Int. J. Biol. Biomed. Med. Sci., vol. Vol. 3, no. No.
3, pp. 1-8, 2008.

[22] A. D. and N. R. Niti Guru, “Decision Support System for Heart Disease Diagnosis
using Neural Network ,” Delhi Bus. Rev., vol. Vol. 8, no. 1, pp. 1–6.

[23] S. P. and R. Awang, “Intelligent Heart Disease Prediction System using Data Mining

105
 Heart Disease Prediction using Machine Learning Algorithms

Techniques,” Int. J. Comput. Sci. Netw. Secur., vol. Vol. 8, no. No. 8, p. pp. 1–6, 2008.

[24] Shantakumar B. Patil and Y.S. Kumaraswamy, “‘Intelligent and Effective Heart Attack
Prediction System using Data Mining and Artificial Neural Network’,” Eur. J. Sci.
Res., vol. Vol. 31, no. No. 4, p. pp. 642-656, 2009.

[25] X. Y. et Al., “Combination Data Mining Models with New Medical Data to Predict
Outcome of Coronary Heart Disease,” Proc. Int. Conf. Converg. Inf. Technol., pp.
868–872, 2007.

[26] E. Y. and C. Kilikcier, “Determination of Patient State from Cardiotocogram using LS-
SVM with Particle Swarm Optimization and Binary Decision Tree,” Master Thesis,
Dep. Electr. Electron. Eng., no. Uludag, 2013.

[27] N. S. and D. Singh, “Performance Evaluation of K-Means and Hierarchal Clustering in

Terms of Accuracy and Running Time,” Dep. Comput. Sci. Eng. Barkatullah Univ.
Inst. Technol., no. Ph.D Dissertation, 2012.

[28] S. G. Manoj B, Kumar G, Ramesh G, “Emerging risk factors for cardiovascular

diseases: Indian context.,” Indian J Endocrinol Metab, vol. 17, pp. 806–814, 2013.

[29] Elama Zannatul F., “Combination of Naive Bayes classifier and K-NN in the
classification based classification models.,” Comput. Inf. Sci, no. 6, pp. 48–56, 2013.

[30] Halaudi Daniel M., “Prediction of heart disease using classification algorithms.,”
WCSECS, pp. 22–24, 2014.

[31] A. Uma ND, “Extraction of action rules for chronic kidney disease using Naive Bayes
classifier.,” IEEE Int Conf. Comput Intell. Comput Res, 2016.

[32] J. A. Marx et al., Rosen’s emergency medicine : concepts and clinical practice, Eighth.
Philadelphia, PA: Elsevier/Saunders, 2014.

[33] M. S., “Integrating Naive and clustering with a different initial centroid selection
methods in the diagnosis of heart disease prediction.,” CS IT CSCP, pp. 125–137,
2012.

[34] Rupali RP MS., “Heart disease prediction system using naive based and Jelmeck
Mercer smoothing.,” IJARCCE, no. 3, pp. 6787–6792, 2014.

[35] C. Aflori, M. Craus, A.J.T. Lee, Y.H. Liu, H.Mu Tsai, H.-Hui Lin, H-W. Wu, “Grid
implementation of the Apriori algorithm Advances in Engineering Software,” Min.

106
 Heart Disease Prediction using Machine Learning Algorithms

Freq. patterns image databases with 9D-SPA Represent. - J. Syst. Softw., vol. Volume
38, no. Issue 5, pp. 295-300.

[36] K. Srinivas, “Analysis of coronary heart disease and prediction of heart attack in coal
mining regions using data mining techniques,” IEEE Trans. Comput. Sci. Educ.
(ICCSE), p. p(1344-1349), 2010.

[37] Y. S. K. Shanta kumar, B.Patil, “Predictive data mining for medical diagnosis of heart
disease prediction,” IJCSE, vol. Vol .17, 2011.

[38] M. Anbarasi et. al., “Enhanced Prediction of Heart Disease with Feature Subset
Selection using Genetic Algorithm,” Int. J. Eng. Sci. Technol., vol. 2, no. 10, pp.
5370–5376, 2010.

[39] H. W. Khaing, “Data Mining based Fragmentation and Prediction of Medical Data,”
IEEE, 2011.

[40] V. Chauraisa and S. Pal, “Data Mining Approach to Detect Heart Diseases,” Int. J.
Adv. Comput. Sci. Inf. Technol., vol. Vol. 2, no. No. 4, p. pp 56-66., 2013.

[41] Q. J., “Induction of decision trees.,” Mach Learn, vol. 1, p. 81—106, 1986.

[42] V. L. Roger et al., “Heart disease and stroke statistics-2011 update: A report from the
American Heart Association,” Circulation, 2011.

[43] R. S. Vasan et al., “Relative importance of borderline and elevated levels of coronary
heart disease risk factors,” Annals of Internal Medicine. 2005.

[44] R. B. D’Agostino et al., “General cardiovascular risk profile for use in primary care:
The Framingham heart study,” Circulation, 2008.

[45] NCEP - National Cholesterol Education Program, “Third Report of the National
Cholesterol Education Program (NCEP) Expert Panel on,” 01-3670, 2001.

[46] S. M. Grundy et al., “Implications of recent clinical trials for the National Cholesterol
Education Program Adult Treatment Panel III Guidelines.,” in Journal of the American
College of Cardiology, 2004.

[47] J. Hippisley-Cox, C. Coupland, Y. Vinogradova, J. Robson, M. May, and P. Brindle,

“Derivation and validation of QRISK, a new cardiovascular disease risk score for the
United Kingdom: Prospective open cohort study,” Br. Med. J., 2007.

107
 Heart Disease Prediction using Machine Learning Algorithms

[48] M. Woodward, P. Brindle, and H. Tunsfall-Pedoe, “Adding social deprivation and

family history to cardiovascular risk assessment: The ASSIGN score from the Scottish
Heart Health Extended Cohort (SHHEC),” Heart, 2007.

[49] R. Jackson, “Updated New Zealand cardiovascular disease risk-benefit prediction

guide,” BMJ, 2000.

[50] X. F. Zhang, J. Attia, C. D’Este, X. H. Yu, and X. G. Wu, “A risk score predicted
coronary heart disease and stroke in a Chinese cohort,” J. Clin. Epidemiol., 2005.

[51] P. M. Ridker, J. E. Buring, N. Rifai, and N. R. Cook, “Development and validation of

improved algorithms for the assessment of global cardiovascular risk in women: The
Reynolds Risk Score,” J. Am. Med. Assoc., 2007.

[52] R. M. Conroy et al., “Estimation of ten-year risk of fatal cardiovascular disease in

Europe: the SCORE project.,” Eur. Heart J., 2003.

[53] I. Graham et al., “European guidelines on cardiovascular disease prevention in clinical

practice: executive summary,” Eur. J. Cardiovasc. Prev. Rehabil., 2007.

[54] A. D. Sniderman and C. D. Furberg, “Age as a modifiable risk factor for

cardiovascular disease,” Lancet, 2008.

[55] D. M. Lloyd-Jones et al., “Prediction of lifetime risk for cardiovascular disease by risk
factor burden at 50 years of age,” Circulation, 2006.

[56] D. F. Terry et al., “Cardiovascular risk factors predictive for survival and morbidity-
free survival in the oldest-old Framingham Heart Study participants,” J. Am. Geriatr.
Soc., 2005.

[57] R. J. Goldberg, M. Larson, and D. Levy, “Factors associated with survival to 75 years
of age in middle-aged men and women: The Framingham study,” Arch. Intern. Med.,
1996.

[58] W. B.J. et al., “Midlife risk factors and healthy survival in men,” J. Am. Med. Assoc.,
2006.

[59] P. S. Yusuf et al., “Effect of potentially modifiable risk factors associated with
myocardial infarction in 52 countries (the INTERHEART study): Case-control study,”
Lancet, 2004.

[60] R. B. D’Agostino et al., “General Cardiovascular Risk Profile for Use in Primary

108
 Heart Disease Prediction using Machine Learning Algorithms

Care,” Circulation, 2008.

[61] R. S. Vasan and R. B. D’Agostino, “Age and Time Need Not and Should Not Be
Eliminated From the Coronary Risk Prediction Models,” Circulation, 2005.

[62] R. M. Epstein, B. S. Alper, and T. E. Quill, “Communicating evidence for

participatory decision making,” Journal of the American Medical Association. 2004.

[63] D. M. Lloyd-Jones et al., “Lifetime risk for developing congestive heart failure: The
Framingham Heart Study,” Circulation, 2002.

[64] D. M. Lloyd-Jones, M. G. Larson, A. Beiser, and D. Levy, “Lifetime risk of

developing coronary heart disease,” Lancet, 1999.

[65] W. B. Kannel and R. S. Vasan, “Is Age Really a Non-Modifiable Cardiovascular Risk
Factor?,” Am. J. Cardiol., 2009.

[66] S. Seshadri et al., “The lifetime risk of stroke: Estimates from the framingham study,”
Stroke, 2006.

[67] J. Canto et al., “Prevalence, clinical characteristics, and mortality among patients with
myocardial infarction presenting without chest pain,” JAMA, vol. 283, no. 24, pp.
3223–9, 2000.

[68] V. L. Roger et al., “Executive summary: Heart disease and stroke statistics-2012
update: A report from the American heart association,” Circulation, 2012.

[69] L. Mosca, H. Mochari-Greenberger, R. J. Dolor, L. K. Newby, and K. J. Robb,

“Twelve-year follow-up of American women’s awareness of cardiovascular disease
risk and barriers to heart health,” Circ. Cardiovasc. Qual. Outcomes, 2010.

[70] R. B. Jaffe, “Explaining the decrease in U.S. deaths from coronary disease, 1980-2000:
Commentary,” Obstetrical and Gynecological Survey. 2007.

[71] J. S. Berger et al., “Sex differences in mortality following acute coronary syndromes,”
JAMA - J. Am. Med. Assoc., 2009.

[72] K. P. Alexander et al., “Sex differences in major bleeding with glycoprotein IIb/IIIa
inhibitors: Results from the CRUSADE (Can rapid risk stratification of unstable
angina patients suppress adverse outcomes with early implementation of the
ACC/AHA guidelines) initiative,” Circulation, 2006.

109
 Heart Disease Prediction using Machine Learning Algorithms

[73] S. O.M. et al., “Improved clinical outcomes in patients undergoing coronary artery
bypass grafting with coronary endarterectomy,” Ann. Thorac. Surg., 1999.

[74] J. D. Puskas, P. D. Kilgo, M. Kutner, S. V Pusca, O. Lattouf, and R. A. Guyton, “Off-

pump techniques disproportionately benefit women and narrow the gender disparity in
outcomes after coronary artery bypass surgery,” Circulation, 2007.

[75] R. P. Hertz, A. N. Unger, J. A. Cornell, and E. Saunders, “Racial disparities in

hypertension prevalence, awareness, and management,” Arch. Intern. Med., 2005.

[76] E. L. Barrett Connor, B. A. Cohn, D. L. Wingard, and S. L. Edelstein, “Why Is

Diabetes Mellitus a Stronger Risk Factor for Fatal Ischemic Heart Disease in Women
Than in Men?: The Rancho Bernardo Study,” JAMA J. Am. Med. Assoc., 1991.

[77] C. C. Cowie et al., “Full accounting of diabetes and pre-diabetes in the U.S. population
in 1988-1994 and 2005-2006,” Diabetes Care, 2009.

[78] R. B. Ervin, “Prevalence of metabolic syndrome among adults 20 years of age and
over, by sex, age, race and ethnicity, and body mass index: United States, 2003-2006.,”
Natl. Health Stat. Report., 2009.

[79] K. M. Flegal, M. D. Carroll, C. L. Ogden, and L. R. Curtin, “Prevalence and trends in

obesity among US adults, 1999-2008,” JAMA - J. Am. Med. Assoc., 2010.

[80] J. R. Pleis, J. W. Lucas, and W. BW, “Summary health statistics for U.S. adults:
National Health Interview Survey, 2008.,” Vital Health Stat. 10., 2009.

[81] L. L. Leape, J. S. Weissman, E. C. Schneider, R. N. Piana, C. Gatsonis, and A. M.

Epstein, “Adherence to practice guidelines: The role of specialty society guidelines,”
Am. Heart J., 2003.

[82] T. M. Shaneyfelt, M. F. Mayo-Smith, and J. Rothwangl, “Are guidelines following

guidelines?. The methodological quality of clinical practice guidelines in the peer-
reviewed medical literature,” J. Am. Med. Assoc., 1999.

[83] M. D. Cabana and C. Kim, “Physician adherence to preventive cardiology guidelines

for women,” in Women’s Health Issues, 2003.

[84] L. Mosca et al., “National study of physician awareness and adherence to

cardiovascular disease prevention guidelines,” Circulation. 2005.

[85] J. Barnhart, V. Lewis, J. L. Houghton, and P. Charney, “Physician Knowledge Levels

110
 Heart Disease Prediction using Machine Learning Algorithms

and Barriers to Coronary Risk Prevention in Women. Survey Results from the Women
and Heart Disease Physician Education Initiative,” Women’s Heal. Issues, 2007.

[86] W. R. Lewis et al., “Trends in the use of evidence-based treatments for coronary artery
disease among women and the elderly: Findings from the get with the guidelines
quality-improvement program,” Circ. Cardiovasc. Qual. Outcomes, 2009.

[87] A. H. Christian, T. Mills, S. L. Simpson, and L. Mosca, “Quality of cardiovascular

disease preventive care and physician/practice characteristics,” J. Gen. Intern. Med.,
2006.

[88] C. L. Bryson, R. R. Miller, A. E. Sales, B. Kopjar, and S. D. Fihn, “Adherence to

Heart-Healthy Behaviors in a Sample of the U.S. Population,” Prev. Chronic Dis.
Electron. Resour., 2005.

[89] T. G. Pickering et al., “Recommendations for blood pressure measurement in humans

and experimental animals. Part 1: Blood pressure measurement in humans: A
statement for professionals from the subcommittee of professional and public
education of the American Heart Association cou,” Hypertension. 2005.

[90] W. B. Kannel, T. Gordon, and M. J. Schwartz, “Systolic versus diastolic blood

pressure and risk of coronary heart disease. The Framingham study,” Am. J. Cardiol.,
1971.

[91] J. Stamler and R. Stamler, “Intervention for the prevention and control of hypertension
and atherosclerotic diseases: United States and international experience,” Am. J. Med.,
1984.

[92] S. S. Franklin, S. A. Khan, N. D. Wong, M. G. Larson, and D. Levy, “Is pulse pressure
useful in predicting risk for coronary heart disease? The Framingham Heart Study,”
Circulation, 1999.

[93] J. L. Izzo, D. Levy, and H. R. Black, “Importance of systolic blood pressure in older
Americans,” Hypertension, 2000.

[94] J. L. Probstfield, “Prevention of stroke by antihypertensive drug treatment in older

persons with isolated systolic hypertension. Final results of the Systolic Hypertension
in the Elderly Program (SHEP),” Annals of Internal Medicine. 1991.

[95] N. Sarwar et al., “Diabetes mellitus, fasting blood glucose concentration, and risk of

111
 Heart Disease Prediction using Machine Learning Algorithms

vascular disease: A collaborative meta-analysis of 102 prospective studies,” Lancet,

2010.

[96] R. F. Gillum and C. T. Grant, “Coronary heart disease in black populations II. Risk
factors,” Am. Heart J., 1982.

[97] D. Steinberg, “Thematic review series: the pathogenesis of atherosclerosis: an

interpretive history of the cholesterol controversy, part III: mechanistically defining
the role of hyperlipidemia.,” J. Lipid Res., 2005.

[98] E. L. M. Barr et al., “Risk of cardiovascular and all-cause mortality in individuals with
diabetes mellitus, impaired fasting glucose, and impaired glucose tolerance: The
Australian Diabetes, Obesity, and Lifestyle Study (AusDiab),” Circulation, 2007.

[99] G. Roglic et al., “The burden of mortality attributable to diabetes: realistic estimates
for the year 2000.,” Diabetes Care, 2005.

[100] A. D. Association, “Diagnosis and Classification of Diabetes Mellitus DEFINITION

AND DESCRIPTION OF DIABETES MELLITUS,” Diabetes Care, 2014.

[101] E. L. M. Barr, E. J. Boyko, P. Z. Zimmet, R. Wolfe, A. M. Tonkin, and J. E. Shaw,

“Continuous relationships between non-diabetic hyperglycaemia and both
cardiovascular disease and all-cause mortality: The Australian Diabetes, Obesity, and
Lifestyle (AusDiab) study,” Diabetologia, 2009.

[102] E. B. Levitan, Y. Song, E. S. Ford, and S. Liu, “Is nondiabetic hyperglycemia a risk
factor for cardiovascular disease? A meta-analysis of prospective studies,” Arch.
Intern. Med., 2004.

[103] M. Tominaga, H. Eguchi, H. Manaka, K. Igarashi, T. Kato, and A. Sekikawa,

“Impaired glucose tolerance is a risk factor for cardiovascular disease, but not
impaired fasting glucose: The Funagata Diabetes Study,” Diabetes Care, 1999.

[104] J. D. Sorkin, D. C. Muller, J. L. Fleg, and R. Andres, “The relation of fasting and 2-h
postchallenge plasma glucose concentrations to mortality: Data from the Baltimore
Longitudinal Study of Aging with a critical review of the literature,” Diabetes Care,
2005.

[105] P. E. Wändell, A. C. Carlsson, and H. Theobald, “The association between BMI value
and long-term mortality,” Int. J. Obes., 2009.

112
 Heart Disease Prediction using Machine Learning Algorithms

[106] M. É. Piché, J. F. Arcand-Bossé, J. P. Després, L. Pérusse, S. Lemieux, and S. J.

Weisnagel, “What is a normal glucose value? Differences in indexes of plasma glucose
homeostasis in subjects with normal fasting glucose,” Diabetes Care, 2004.

[107] M. A. Westwood et al., “Normalized left ventricular volumes and function in

thalassemia major patients with normal myocardial iron,” J. Magn. Reson. Imaging,
2007.

[108] S. Claster et al., “Nutritional deficiencies in iron overloaded patients with

hemoglobinopathies,” Am. J. Hematol., 2009.

[109] J. C. Wood et al., “Vitamin D deficiency, cardiac iron and cardiac function in
thalassaemia major,” Br. J. Haematol., 2008.

[110] A. El-Beshlawy et al., “Improvement of cardiac function in thalassemia major treated

with L-carnitine,” Acta Haematol., 2004.

[111] J. C. Wood et al., “Physiology and pathophysiology of iron cardiomyopathy in

thalassemia,” in Annals of the New York Academy of Sciences, 2005.

[112] U. N. Dulhare and M. Ayesha, “Extraction of action rules for chronic kidney disease
using Naïve bayes classifier,” in 2016 IEEE International Conference on
Computational Intelligence and Computing Research, ICCIC 2016, 2017.

[113] T. J. Peter and K. Somasundaram, “Study and Development of Novel Feature

Selection Framework for Heart Disease Prediction,” Int. J. Sci. Res. Publ., 2012.

[114] B. Xue, M. Zhang, and W. N. Browne, “Particle swarm optimization for feature
selection in classification: A multi-objective approach,” IEEE Trans. Cybern., 2013.

[115] Ks. BKavihta Rani AGovrdhan Associate Professor, P. of CSE, and N. Karimnagar
Jagtial, “Applications of Data Mining Techniques in Healthcare and Prediction of
Heart Attacks,” 2010.

[116] N. A. Setiawan, P. A. Venkatachalam, and A. F. M. H, “Rule Selection for Coronary

Artery Disease Diagnosis Based on Rough Set,” Int. J. Recent Trends Eng. Technol.,
2009.

[117] F. Thabtah, “A review of associative classification mining,” Knowledge Engineering

Review. 2007.

[118] F. Tao, F. Murtagh, and M. Farid, “Weighted Association Rule Mining using weighted

113
 Heart Disease Prediction using Machine Learning Algorithms

support and significance framework,” 2004.

[119] M. Stensmo and T. J. Sejnowski, “Automated Medical Diagnosis based on Decision

Theory and Learning from Cases,” World Congr. Neural Networks, 1996.

[120] Wenmin Li, Jiawei Han, and Jian Pei, “CMAR: accurate and efficient classification
based on multiple class-association rules,” 2002.

[121] M. Al Mamun et al., “Emerging Burden of Cardiovascular Diseases in Bangladesh.”

[122] M. Joel M. Gore, “Typical Angina vs. Atypical Chest Pain,” NEJM J. Watch, vol.
2010, Jul. 2010.

[123] J. Constant, “The diagnosis of nonanginal chest pain.,” Keio J. Med., vol. 39, no. 3, pp.
187–92, Sep. 1990.

[124] B. Unal, J. A. Critchley, and S. Capewell, “Modelling the decline in coronary heart
disease deaths in England and Wales, 1981-2000: comparing contributions from
primary prevention and secondary prevention,” BMJ, vol. 331, no. 7517, p. 614, Sep.
2005.

[125] “Diabetes - Diagnosis and treatment - Mayo Clinic.” [Online]. Available:

https://www.mayoclinic.org/diseases-conditions/diabetes/diagnosis-treatment/drc-
20371451. [Accessed: 28-Apr-2019].

[126] “Angina (Chest Pain) | American Heart Association.” [Online]. Available:

https://www.heart.org/en/health-topics/heart-attack/angina-chest-pain. [Accessed: 28-
Apr-2019].

[127] H. L. Fred, “Atypical chest pain: a typical humpty dumpty coinage.,” Texas Hear. Inst.
J., vol. 36, no. 5, pp. 373–4, 2009.

[128] J. Robson, L. Ayerbe, R. Mathur, J. Addo, and A. Wragg, “Clinical value of chest pain
presentation and prodromes on the assessment of cardiovascular disease: a cohort
study.,” BMJ Open, vol. 5, no. 4, p. e007251, Apr. 2015.

[129] D. Pei et al., “Relationship of Blood Pressure and Cardiovascular Disease Risk Factors
in Normotensive Middle-Aged Men,” Medicine (Baltimore)., vol. 90, no. 5, pp. 344–
349, Sep. 2011.

[130] “How to read an Electrocardiogram (ECG). Part One: Basic principles of the ECG.
The normal ECG.” [Online]. Available:

114
 Heart Disease Prediction using Machine Learning Algorithms

http://www.southsudanmedicaljournal.com/archive/may-2010/how-to-read-an-
electrocardiogram-ecg.-part-one-basic-principles-of-the-ecg.-the-normal-ecg.html.
[Accessed: 28-Apr-2019].

[131] R. E. O’Connor et al., “Part 10: acute coronary syndromes: 2010 American Heart
Association Guidelines for Cardiopulmonary Resuscitation and Emergency
Cardiovascular Care.,” Circulation, vol. 122, no. 18 Suppl 3, pp. S787-817, Nov.
2010.

[132] M. Kukar, I. Kononenko, C. Grošelj, K. Kralj, and J. Fettich, “Analysing and

improving the diagnosis of ischaemic heart disease with machine learning,” Artif.
Intell. Med., vol. 16, no. 1, pp. 25–50, May 1999.

115