Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165 [Link]

Heart Disease Detection Using AI

Narannagari Chaathurya (Student) Sikharam Abhinav (Student)
Artificial Intelligence and Machine Learning Artificial Intelligence and Machine Learning
Sphoorthy Engineering College (JNTUH) Sphoorthy Engineering College (JNTUH)
Hyderabad, India Hyderabad, India

Battu Sri Vamshidhar (Student) Kandula Revathi (Assistant professor)

Artificial Intelligence and Machine Learning Artificial Intelligence and Machine Learning
Sphoorthy Engineering College (JNTUH) Sphoorthy Engineering College (JNTUH)
Hyderabad, India Hyderabad, India

Abstract:- Over the past few decades, cardiovascular history, and diagnostic test results, to provide valuable
disease has emerged as the primary cause of death insights to healthcare professionals.
worldwide in both industrialized and developing nations.
Early detection of heart problems and continued clinical The use of Python in healthcare extends beyond just
monitoring can reduce death rates. However, because it predicting heart diseases; it also facilitates the development
takes more time and experience, it is not possible to of applications for managing patient records, analysing
accurately detect heart disorders in all cases and to have medical imaging data, and even assisting in surgical
a specialist talk with a patient for 24 hours. We procedures. The language's ease of use and extensive
demonstrate how machine learning can be used to community support contribute to its widespread adoption in
estimate an individual's risk of developing heart disease. the medical field.
This study presents data processing, which includes
converting categorical columns and working with Overall, the combination of machine learning
categorical variables. We outline the three primary techniques and Python programming offers promising
stages of developing an application: gathering datasets, opportunities for improving healthcare outcomes,
running logistic regression, and assessing the properties particularly in the early detection and management of heart
of the dataset. The random forest classifier technique is diseases.
developed to diagnose cardiac problems more precisely.
Data analysis is needed for this application since it is II. PROBLEM STATEMENT
considered noteworthy. The random forest classifier
algorithm, which improves the accuracy of research It's evident that the healthcare sector is increasingly
diagnosis, is next covered, along with the experiments relying on data-driven insights to improve patient care and
and findings. optimize healthcare delivery. Python, with its versatility and
extensive libraries, plays a crucial role in extracting valuable
Keywords:- Artificial Intelligence; Early Detection; insights from healthcare data.
Machine Learning; Heart Disease Detection; Data Analysis.
Regarding heart disease, Python can aid in analysing
I. INTRODUCTION various factors such as cholesterol levels, patient
demographics, and medical history to predict and diagnose
This paper discusses the relevance of Python conditions like coronary artery disease (CAD). As
programming language in healthcare applications, mentioned, CAD often goes undetected in its early stages,
specifically in the development of dynamic and scalable making predictive analytics especially valuable for early
solutions for heart disease detection, and the significance of intervention.
machine learning in cardiac disease diagnosis and
prediction. Python's versatility and rich ecosystem of Moreover, Python's capabilities extend to ensuring
libraries indeed make it a popular choice for such tasks. compliance with regulations like HIPAA, which are
paramount in handling sensitive healthcare records. With
In the context of heart diseases, machine learning built-in tools for software-defined security, Python helps
models can analyse various medical data to predict the healthcare projects adhere to strict data protection standards.
likelihood of a patient having a heart condition. By
leveraging libraries like Pandas for data manipulation, Machine learning algorithms further enhance
Matplotlib for data visualization, developers can build healthcare analytics by enabling the development of tracking
robust predictive models. These models can process diverse and health monitoring applications. Python's ease of use and
data sources, including patient demographics, medical robust libraries make it an ideal choice for building these
applications, ultimately leading to better patient outcomes.

IJISRT24MAR211 [Link] 227

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 [Link]

Python’s prominence in the healthcare sector stems healthcare solutions, including those aimed at detecting and
from its ability to handle complex data analysis tasks, ensure managing heart disease.
data security, and facilitate the development of innovative

III. LITERATURE REVIEW

Fig. 1. Theoretical Framework.

It's important to understand the risk factors associated Chol, fbs, and others. Libraries such as NumPy, pandas,
with heart disease, as well as the symptoms that may matplotlib and scikit-learn were used.
indicate a heart problem. Diabetes, obesity, unhealthy diet,
overweight, excessive alcohol use, and physical inactivity Machine learning classifiers such as K Neighbors
are all significant contributors to heart disease risk. And Classifier, Random Forest Classifier, Logistic Regression,
while chest pain is a common symptom and often a warning and Decision Tree Classifier can be applied to predict heart
sign of cardiovascular issues, it's important to note that disease based on input features. These algorithms can learn
symptoms like nausea, indigestion, heartburn, or stomach from historical data to classify new instances into different
pain can also sometimes be associated with heart problems, categories, such as presence or absence of heart disease.
particularly in women. It's always crucial to pay attention to Hybrid methods, involve combining multiple algorithms or
any unusual symptoms and consult a healthcare professional techniques to improve the accuracy and robustness of the
if one has concerns about heart health. predictive models. This could include integrating logistic
regression, K-nearest neighbor, and neural networks to
Additionally, employing machine learning techniques develop more comprehensive heart disease diagnostic
can assist in diagnosing and predicting heart disease based algorithms. It's an exciting and important area of research
on the relevant features in the dataset and patient and application, as artificial intelligence and machine
information. Using a correlation matrix can help identify learning can potentially enhance medical diagnosis and
relationships between different variables related to heart treatment by leveraging large datasets and advanced
disease, while histograms can provide insights of algorithms to extract meaningful insights.
distribution of these variables within the dataset. The dataset
comprises of several factors, such as, age, sex, cp, trestbps,

IJISRT24MAR211 [Link] 228

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 [Link]

IV. METHODOLOGY G. Model Interpretation and Validation:

 Interpret the trained models to understand their decision-
Developing heart disease detection using AL involves making process and identify important features for heart
a systematic methodology to ensure the model's disease detection.
effectiveness, reliability, and ethical considerations. Here's a  Validate the models with domain experts to ensure their
step-by-step methodology: clinical relevance and interpretability.

A. Objectives V. PROPOSED SYSTEM

Critically evaluate the methods used to acquire and
pre-process the heart disease data. Assess the quality of the A. Description of Suitable Libraries:
data sources, including their reliability, completeness, and Python's popularity in the programming community is
representativeness. well-deserved, especially in the realm of data science and
machine learning. Let's delve into the purposes of the
Evaluating the effectiveness of feature selection libraries you mentioned:
methods in identifying informative features and reducing
dimensionality while preserving predictive power.  Numpy:
This library is fundamental for numerical computing in
Critically assessing the reliability and generalization Python. Large, multi-dimensional arrays and matrices are
capability of the models. supported, and a number of mathematical operations are
available for effective manipulation of these arrays. The
B. Data Collection and Exploration: foundation of many other libraries in the ecosystem of
Gather relevant datasets containing features related to scientific computing is NumPy.
heart disease, such as demographic information, medical
history, and diagnostic test results. Explore the data to  Pandas:
understand its characteristics, identify potential biases or Pandas is a powerful data manipulation and analysis
missing values, and gain insights into feature distributions library. It provides data structures like Data Frame and
and relationships. Series that make it easy to work with structured data,
perform data cleaning, manipulation, and analysis tasks.
C. Data Pre-processing: pandas is particularly useful for handling tabular data, such
 Handle missing numbers, outliers, and inconsistent data as CSV files or SQL database tables.
to make the data cleaner.
 Encode categorical variables and perform feature scaling  Scikit-Learn:
or normalization as necessary. Scikit-learn is one of the most popular machine
 Split the data into training, validation, and testing sets to learning libraries in Python. It offers a wide range of
facilitate model training and evaluation. supervised and unsupervised learning algorithms, along with
tools for model selection, evaluation, and pre-processing.
D. Feature Selection and Engineering: scikit-learn's user-friendly interface makes it accessible for
 Select informative features that are likely to be both beginners and experts in machine learning.
predictive of heart disease based on domain knowledge
and data exploration.  Matplotlib:
 Engineer new features or transformations to enhance the Matplotlib is a plotting library for creating static,
predictive power of the model. interactive, and animated visualizations in Python. It
provides a MATLAB-like interface for creating plots and
E. Model Selection: charts, making it easy to generate a wide variety of graphical
 Choose appropriate ML algorithms for heart disease representations of data. matplotlib is often used in
detection based on the nature of the problem, data combination with NumPy and pandas for visualizing data
characteristics, and computational resources. and analysis results.
 Consider a variety of algorithms, such as logistic
regression, decision trees, random forest, support vector  Seaborn:
machines, neural networks, or ensemble methods. For Python statistical graphics plotting, Seaborn is an
incredible visualization library. In order to enhance the
F. Model Training and Evaluation: visual appeal of statistical graphs, it offers lovely default
 Train the selected ML models using the training data and styles and color schemes. The data structures from Pandas
validate their performance using the validation set. are strongly interwoven with the matplotlib library upon
 Evaluate the models using appropriate evaluation which it is developed.
metrics, such as accuracy, precision, recall, F1-score,
area under the ROC curve, or confusion matrix. These libraries, when combined, form a powerful
 Perform cross-validation to assess the models' toolkit for data scientists and machine learning practitioners,
generalization performance and robustness. enabling them to efficiently explore, manipulate, analyse,
and model data, as well as visualize their findings. Their
versatility and extensive documentation make them

IJISRT24MAR211 [Link] 229

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 [Link]

indispensable assets in the field of data science and artificial  K-Nearest Neighbor:
intelligence. KNN (k-Nearest Neighbors) is utilized in heart disease
detection by classifying patients based on the majority class
B. Description of Suitable ML Algorithms: of their nearest neighbors. It is a non-parametric, lazy
Using Machine Learning algorithms like Random learning algorithm that classifies data points based on the
Forest, k-Nearest Neighbors, Decision tree Classifier, majority class among their nearest neighbors.
utilized to classify heart disease risk. Correlation matrix
analysis helps identify the relationships between different  Logistic Regression:
variables and heart disease indicators. Logistic regression is a type of regression analysis used
for predicting the probability of a binary outcome (such as
 Decision Tree: the presence or absence of heart disease) based on one or
Decision tree algorithms are employed in heart disease more predictor variables. It is employed in heart disease
detection to create predictive models based on splitting data detection to model the probability of patients having heart
into hierarchical decision nodes. Decision trees are intuitive disease based on their characteristics.
models that recursively split the data based on features,
aiming to create homogeneous subsets that are more  Random Forest:
predictive of the target variable—in this case, the presence The random forest algorithm is utilized in heart disease
or absence of heart disease. detection to build an ensemble of decision trees, where each
tree is trained on a random subset of the data and votes on
the final classification, providing robustness and accuracy in
prediction. It's often used for classification tasks, including
heart disease detection.

Fig. 2. Random Forest Algorithm.

VI. DATA DESCRIPTION One crucial step in the pre-processing of data is

dividing the dataset into train and test sets. This allows us to
The following URL contains the dataset used to enhance the predictability of our model and increase its
identify and analyse heart diseases: performance. It is possible to think of it this way: if we use a
[Link] training set to train our model and an entirely different test
dataset to test it on, our model will be unable to recognize
Age, sex, cp, trestbps, chol, fbs, restecg, thalach, the correlations between the [Link], the
exang, oldpeak, slope, ca, thal, and num are among the model's performance will suffer if we use two different
variables that are employed. In this project, a total of 14 datasets for training and testing. Therefore, it is crucial to
factors that can improve the accuracy of heart disease divide a dataset into a train set and a test set. This makes it
detection are taken into consideration. This dataset contains simple for us to assess how well our model is performing.
14 columns of sample data from 302 patients. For the For example, if the model performs well on training data but
purpose of detecting heart disease, every trait is equally poorly on the test dataset, it's possible that the model is
important. The data is split between 70% and 30% of the overfitted.
303 samples. A total of 212 samples i.e., 70% of the data,
are used to train the model. The remaining 91 samples i.e., The train_test_split function from scikit-learn can be
30% of the data, are used to test the model. used to split the dataset.

IJISRT24MAR211 [Link] 230

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 [Link]

 Training Data: clinicians and institutions to improve patient outcomes

The first set of data used to build a machine learning through scalable and dynamic applications. Chapter two
model—from which the model builds and improves its further delves into the application of Python for detecting
rules—is referred to as training data. The quality of this data the presence of heart diseases, utilizing a dataset containing
has a significant impact on how the model is developed patient data such as age, sex, cholesterol levels (Chol), and
going forward, establishing a strong standard for all other relevant factors.
applications that use the same training set in the future.
Individual libraries such as Matplotlib, NumPy,
 Testing Data: Pandas, warnings, and others are imported for use in the
The model needs to be tested using the test dataset heart disease detection application. Python, being a robust
after it has been trained using the training dataset. Using the language, is noted for its computational capabilities,
new or unknown dataset, this dataset assesses the model's facilitating the extraction of valuable insights from patient
performance and guarantees that it can generalize information related to heart diseases. Additionally, it is
successfully. A different subset of the original data that is emphasized that Python complies with HIPAA regulations,
unrelated to the training dataset is the test dataset. ensuring the safety of medical information. Machine
learning is underscored as crucial for predicting threats like
As is customary, the patient's age is recorded in years, heart disease. Specifically, the Random Forest algorithm is
their sex is recorded as 0/1, with 1 being a male patient and chosen for developing the heart disease detection
0 representing a female patient. The type of chest pain (CP) methodology. Furthermore, it's mentioned that the ML
is recorded as 1, 2, 3, and 4. One for typical angina pain, two model, particularly Random Forest classification, plays a
for atypical angina, three for non-anginal pain, and four for significant role in achieving accuracy and determining
asymtomatic. Upon admission to the hospital, the patient's results using training data. The selection of Random Forest
resting blood pressure (trestbps) is measured in millimeter- is based on the specific dataset, as well as its comparison
Hg. Chol is measured in milligrams per deciliter of serum with the decision tree algorithm. Data analysis is
cholesterol. The fasting blood sugar, or fbs, is entered as 0/1 highlighted, particularly the handling of categorical
and should be greater than 120 mg/dl. 1 denotes true and 0 variables by breaking them into dummy columns with
falsehood. Resting electrocardiogram (resting ECG) findings binary values (1s and 0s). The output of the application
are entered as 0/1/2, where 0 indicates normal, 1 indicates includes medical parameters such as age, gender, blood
ST-T wave abnormalities (T wave inversions and/or ST pressure, cholesterol, and obesity, which are used for
elevation or depression of > 0.05 mV), and 2 indicates prediction and software development requirements. Machine
probable or definitive left ventricular hypertrophy according learning application using Python is described as a subset of
to Estes' criteria. The highest heart rate attained is thalach. the Artificial Intelligence model, with Python libraries being
Exang is the angina exercise that accepts input values of 0/1. essential for making predictions. The Scikit-learn
0 means no and 1 means yes. Oldpeak, as opposed to rest, is (SKLEARN) library is specifically mentioned as commonly
the ST depression brought on by exertion. used in machine learning prediction tasks. Random Forest is
identified as the preferred algorithm for predicting heart
Slope, which accepts 1/2/3 as input, is the slope of the peak disease due to its simplicity and ability to produce precise
exercise ST segment. Three for downsloping, two for flat, results.
and one for upsloping. Approximately for the number of
large vessels (0–3), colored by flourosopy. The thalasemia REFERENCES
indicator, or thal, has three values: 3 for normal, 6 for fixed
defects, and 7 for reversible defects. The diagnosis of [1]. L. Loku, B. Fetaji, A. Krstev, M. Fetaji, Z. Zdravev,
cardiac illness, num or target, accepts an input of 0/1. A Using python programming for assessing and solving
diameter narrowing of 0 means less than 50%, whereas a health management issues, South East Eur. J.
narrowing of 1 means more than 50%. [Link]. 4 (1) (2020).
[Link]
VII. CONCLUSION [2]. P. Guleria, M. Sood, Intelligent learning analytics in
healthcare sector using machine learning, in: Machine
The discussion revolves around utilizing Python for Learning with Health Care Perspective,
heart disease prediction and detection. Python is highlighted Springer,Cham, 2020.
as an object-oriented, high-level programming language [Link]
with quick development cycles and robust building options. 40850-3_3
This language is deemed beneficial for accurately predicting [3]. Spencer R., Thabtah F., Abdelhamid N., Thompson M.
the pathway of heart disease due to its attributes. The Exploring feature selection and classification methods
healthcare industry, particularly the heart care sector, is for predicting heart disease. Digital Health . 2020.
emphasized as actively generating data from various [Link]
facilities and patients, leveraging effective data strategies. 914777
Additionally, doctors are depicted as utilizing superior
predictive models for treatments, thereby enhancing the
overall healthcare delivery system. The prediction model for
heart disease is specifically mentioned as being employed by

IJISRT24MAR211 [Link] 231

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 [Link]

[4]. Javeed A., Zhou S., Yongjian L., Qasim I., Noor A., [13]. D. Pedrozo, F. Barajas, A. Estupiñán, K.L. Cristiano,
Nour R. An intelligent learning system based on D.A. Triana, Data analysis for a set of university
random search algorithm and optimized random forest student lists using the k-Nearest Neighbors machine
model for improved heart disease detection. IEEE learning method, J. Phys. Conf. Ser. 1514 (1) (2020)
Access . 2019. 1–8. [Link]
[Link] 6596/1514/1/012011/meta
[5]. Drożdż, K.; Nabrdalik, K.; Kwiendacz, H.; Hendel, M.; [14]. A. Anees, I. Hussain, A novel method to identify initial
Olejarz, A.; Tomasik, A.; Bartman, W.; Nalepa, J.; values of chaotic maps in cybersecurity, Symmetry 11
Gumprecht, J.; Lip, G.Y.H. Risk factors for (2) (2019) 140. [Link]
cardiovascular disease in patients with metabolic- 8994/11/2/140
associated fatty liver disease: A machine learning [15]. Y. Fan, J. Li, D. Zhang, J. Pi, J. Song, G. Zhao,
approach. Cardiovasc. Diabetol. 2022. Supporting sustainable maintenance of substations
[Link] under cyber-threats: An evaluation method of
01672-9 cybersecurity risk for power CPS, Sustainability 11 (4)
[6]. Ouf, S.; ElSeddawy, A.I.B. A proposed paradigm for (2019) 1–30. [Link]
intelligent heart disease prediction system using data 1050/11/4/982
mining techniques. J. Southwest Jiaotong Univ. 2021.
[Link]
[7]. A. Zahariev, M. Zveryakov, S. Prodanov, G.
Zaharieva, P. Angelov, S. Zarkova, M. Petrova, Debt
management evaluation through support vector
machines: on the example of Italy and Greece,
Entrepreneurship Sustain. Issues 7 (3) (2020) 1–12.
[Link]
0300483X17303451
[8]. Bhunia, P.K.; Debnath, A.; Mondal, P.; D E, M.;
Ganguly, K.; Rakshit, P. Heart Disease Prediction
using Machine Learning. Int. J. Eng. Res.
Technol. 2021.
[Link]
+Disease+Prediction+using+Machine+Learning&auth
or=Bhunia,+P.K.&author=Debnath,+A.&author=Mond
al,+P.&author=D+E,+M.&author=Ganguly,+K.&autho
r=Rakshit,+P.&publication_year=2021&journal=Int.+J
.+Eng.+Res.+Technol.&volume=9
[9]. Hassan, C.A.U.; Iqbal, J.; Irfan, R.; Hussain, S.;
Algarni, A.D.; Bukhari, S.S.H.; Alturki, N.; Ullah, S.S.
Effectively Predicting the Presence of Coronary Heart
Disease Using Machine Learning
Classifiers. Sensors 2022.
[Link]
[10]. Subahi, A.F.; Khalaf, O.I.; Alotaibi, Y.; Natarajan, R.;
Mahadev, N.; Ramesh, T. Modified Self-Adaptive
Bayesian Algorithm for Smart Heart Disease
Prediction in IoT System. Sustainability 2022.
[Link]
[11]. P. Mathur, Overview of machine learning in
healthcare, in: Machine Learning Applications using
Python, A Press, Berkeley, CA, 2019.
[Link]
3787-8_1
[12]. A. Navlani, Understanding random forests classifier in
python, DataCamp (2018) Available at:
[Link]
m-forestsclassifier-python, [Accessed on 5th March,
2021]. [Link]
forests-classifier-python

IJISRT24MAR211 [Link] 232