Enhancing Accuracy in Heart Disease Prediction: A Hybrid Approach
Enhancing Accuracy in Heart Disease Prediction: A Hybrid Approach
Enhancing Accuracy in Heart Disease Prediction: A Hybrid Approach
ISSN: 1671-5497
E-Publication: Online Open Access
Vol: 42 Issue: 11-2023
DOI: 10.5281/zenodo.10203369
ANKIT SHARMA
Student, M.Tech CSE, Department of CSE, Rayat-Bahra University, Mohali, Punjab, India.
Email: [email protected]
ER. SUKHJINDER KAUR
Department of CSE, Rayat-Bahra University, Mohali, Punjab, India. Email: [email protected]
ER. POONAM KUKANA
Department of CSE ,Rayat-Bahra University, Mohali, Punjab, India. Email: [email protected]
Dr. PUNEET SAPRA
Department of CSE, Rayat-Bahra University, Mohali, Punjab, India. Email: [email protected]
Abstract
Predicting the onset of heart disease accurately is essential for early diagnosis and prevention of this global
pandemic. The paper suggests a hybrid method to improve heart disease prediction. The research
examines several machine learning (ML) models for detecting heart illness and assesses how well they
predict heart disease. To enhance precision, the hybrid method employs not one but many machine
learning methods. The hybrid method employs SVMs, random forests, and neural networks as its machine-
learning algorithms. When it comes to classification, SVM is a very effective method. The data points are
separated into classes, and the optimal hyperplane to do this is the goal. SVM can learn the boundaries
and patterns between various risk variables and efficiently categorize people as having heart disease or
not. Random forests are a kind of ensemble learning that uses several individual decision trees to make a
final determination. The characteristics used to construct each decision tree are chosen at random. Each
decision tree contributes to the overall forecast, which is then aggregated. Due to their versatility, random
forests may be used to the prediction of cardiovascular disease. Neural networks are a kind of algorithm
that takes their cues from the way the human brain operates. They are made up of several layers of artificial
neurons working together to learn intricate patterns from data. Medical diagnosis is only one field where
neural networks have been shown to be useful. In the hybrid method, neural networks may learn complex
associations between risk factors and cardiovascular disease and provide reliable prognoses based on this
information. The hybrid method enhances the accuracy of heart disease prediction by combining the
benefits of various machine-learning techniques
Keywords: Machine Learning, Heart Disease, Feature Learning, Hybrid Approach, Prediction Accuracy,
Ensemble Learning, Performance Measures.
1. INTRODUCTION
Diseases affecting the heart and blood arteries are often referred to as cardiovascular
disease. It places a heavy load on public health systems all around the globe and is a
primary cause of death. Coronary artery disease, congestive heart failure, arrhythmias,
and valvular heart disease are only few of the many kinds of heart disease. Diagnosis,
treatment, and management of each subtype may be difficult. Diseases of the coronary
arteries, heart attacks, heart failure, and irregular heartbeats all fall within this category.
It may be difficult to assess an individual's vulnerability to these illnesses because of the
complex interconnections between genetic, behavioral, and environmental factors at play
in each case. Simple clinical measurements and statistical analysis are widely used in
traditional risk assessment approaches, but they might overlook subtle patterns and
nuances in the development of heart disease.[1]
The most prevalent kind of heart illness is called coronary artery disease (CAD), and it
happens when plaque builds up in the arteries that carry blood to the heart and narrows
or blocks them. Angina chest discomfort, a heart attack, or anything much worse might
result from this. When the heart is unable to pump blood adequately, a condition known
as congestive heart failure (CHF) sets in. This condition causes fluid to accumulate in the
lungs and other regions of the body. Valvular heart disease, in which the heart's valves
are damaged or dysfunctional, restricts blood flow, and arrhythmias, in which the heart's
electrical system is aberrant, both cause irregular heartbeats.
There are several potential causes and contributors to heart disease. There are both
modifiable risk factors, which can be managed, and non-modifiable risk factors, which
cannot be altered in any way. Hypertension, high cholesterol, smoking, obesity, inactivity,
diabetes, and poor nutrition are all preventable risk factors. We can't change the reality
that age, gender, family history, and genetic susceptibility are all risk factors.
The emergence of machine learning-based techniques as powerful tools for analyzing
vast amounts of patient data and extracting actionable insights for predictive modeling is
a relatively recent development. By utilizing the strengths of these algorithms,
researchers have made substantial progress toward better-diagnosing heart illness.
Nevertheless, in spite of their achievements, machine learning models by themselves
sometimes have limits when it comes to capturing the intricacy of the illness. It is possible
that they may not adequately integrate domain-specific information or take into account
the complicated linkages between the different clinical variables that contribute to the risk
of developing heart disease. [2]
Because cardiovascular disease is still one of the main causes of death throughout the
globe, it is essential that accurate prediction models be developed so that those who are
at risk may be identified. The management of this prevalent health problem requires an
accurate diagnosis and prompt intervention in order to be successful. This enables
healthcare practitioners to give focused preventative measures and customized treatment
regimens to patients. In recent years, there has been significant progress made in
increasing the accuracy of predicting cardiac disease by combining medical knowledge
with machine learning algorithms. These latest efforts have shown encouraging
outcomes. To address this challenge, researchers have turned to machine learning
algorithms to improve heart disease prediction. These algorithms can analyze large
datasets and identify complex patterns that may not be apparent through traditional
methods. Support vector machines, random forests, and neural networks are among the
popular algorithms used for heart disease prediction. Each algorithm has its strengths
and weaknesses, and combining them in a hybrid model has the potential to enhance
prediction accuracy.
To address these issues and enhance the accuracy with which heart disease may be
predicted, the authors of this study advise using a hybrid approach. The hybrid model
draws upon the strengths of existing prediction algorithms while also factoring in expert
clinical knowledge and domain-specific data. This strategy tries to give a more robust and
accurate prediction framework by going further than the limits of individual algorithms. It
does this by putting together the parts of different models into a single structure.
In the research paper, there will be information about how the hybrid strategy was made
and how it was used, like which predictive models, clinical features, and machine learning
methods were chosen and how they were put together. In addition, the paper will describe
the training and assessment datasets as well as the evaluation criteria used to the
findings of the hybrid model.
With the help of this study's results, it will be easier to figure out who is at high risk of
getting heart disease and what steps to take to prevent it. This will add to the ongoing
efforts to make heart disease predictions more accurate. If the results of these studies
are used to make personalized treatment plans for people with heart disease, they might
have better disease management and better clinical outcomes.
In conclusion, the purpose of this study is to provide a hybrid method that integrates many
predictive models, clinical features, and cutting-edge machine-learning techniques to
improve the accuracy of heart disease prediction. In order to provide a more
comprehensive framework for precise risk assessment in the context of cardiovascular
disease, the proposed hybrid model aims to overcome the limitations of individual
algorithms. To achieve this goal, we will integrate elements from several strategies.
2. REVIEW OF LITERATURE
Accurate risk assessment techniques are constantly needed since cardiovascular
disease is still a significant public health problem. Researchers have been looking at a
hybrid approach that blends several predictive models with cutting-edge machine learning
methods in recent years to better forecast the likelihood of heart disease. The purpose of
this literature review is to provide a synopsis of the most important studies and research
articles that have explored how a hybrid approach might enhance prediction accuracy.
The research articles and studies will come from a wide range of places, both academic
and professional.
Ali, T., Rahman et al. (2022). The purpose of this work is to give an extensive evaluation
of hybrid machine-learning algorithms for the prediction of heart disease. It investigates
a wide variety of hybrid models, including ensemble approaches, feature selection
strategies, and combination models, emphasizing both the merits and limits of these
models in terms of enhancing the accuracy of predictions. The authors analyze the
effectiveness of these models in a variety of heart disease prediction situations, giving
academics and practitioners insights that might help them pick acceptable hybrid
strategies for their particular requirements. [3]
Poudel, U., & Kim, H. (2022). The purpose of this review article is to give an in-depth
assessment of hybrid machine-learning algorithms for the prediction of heart disease.
This paper investigates the use of hybrid modeling to include a variety of distinct
algorithms, approaches for feature selection, and learning ensembles into the modeling
process. The authors examine the advantages and disadvantages associated with the
use of hybrid approaches and emphasize the relevance of selecting the appropriate
features and algorithms in order to improve the accuracy of prediction. Academics and
medical professionals interested in developing effective hybrid models for cardiac disease
prediction will find this review to be an essential resource. [4]
Xu, L., Zhang, L., Huang, K., & Chen, Z. (2022). With this study, we want to give a
thorough evaluation of hybrid models for heart disease prognosis. The authors undertake
a thorough literature review of several studies and provide a streamlined overview of the
salient characteristics, methodologies, and outcomes of several different hybrid models.
They look at how different algorithms, feature selection methods, and data sources may
be integrated into these models. The systematic review's summary of current state-of-
the-art hybrid approaches for predicting heart disease is a good way to find areas where
more research is needed. [5]
Islam, M. M., et al (2023). This review paper looks at how hybrid deep-learning models
can be used to predict heart disease. It looks into whether deep learning techniques could
be used with other algorithms like decision trees, support vector machines, and random
forests to make better predictions. The good and bad of using hybrid models are laid forth,
with backing given to things like feature extraction, model interpretability, and dealing with
unbalanced data. The feasibility of using hybrid deep-learning models for precise cardiac
disease prediction is investigated. [6]
Han, J., et al (2023). This extensive study aims to use hybrid machine learning algorithms
to EHR-based cardiovascular disease predictions. The authors do an extensive literature
review on a wide range of studies that utilize EHRs in collaboration with various machine
learning techniques. They talk about the pros and cons of using EHR data in hybrid
models, as well as how to choose which features to use, how to prepare the data, and
how to measure the model's performance. Researchers and clinicians interested in using
EHR data to forecast cardiac illness might benefit from the study's guidelines and
suggestions.[7]
Majumder, K., Ghosh, A., GAO, H., & Qiu, M. (2023) the major focus of this review paper
is on using hybrid models to the prediction of heart disease by making use of imbalanced
data. To address the issues brought on by imbalanced datasets, the authors investigate
a number of hybrid approaches, such as cost-sensitive learning, data resampling, and
ensemble methods. The authors also talk about the problems that come up when
datasets are not balanced. They emphasize the performance of hybrid models in
managing unbalanced data and give insights into the strengths and limits of these models
as well as highlight the performance of hybrid models in handling imbalanced data. This
review is a useful resource for academics who are interested in constructing accurate
models for predicting heart disease from unbalanced datasets.[8]
Xie, Y., Li, H., Sun, Y., & Guo, W. (2023). In this extensive study, the topic of hybrid
machine learning models for predicting heart disease based on EHR is the primary
emphasis. Deep learning, support vector machines, and random forests are just some of
examples of hybrid models that the authors investigate. These models merge EHR data
with machine learning methods. They explore the benefits and difficulties associated with
using EHR data in hybrid models, such as data quality, feature extraction, and
interpretability. This study sheds light on the existing state-of-the-art methodologies and
draws attention to the potential of hybrid models that make use of EHR data for the
accurate prediction of cardiac disease. [9]
Wang et al. (2020) developed a hybrid model that combined genetic risk scores with
clinical risk variables and suggested this model. When compared to models that used
genetic variables or clinical factors separately, the accuracy of the combined model was
much higher. A wider variety of risk variables could be captured by the hybrid technique
since it took into account both genetic predisposition and clinical features. As a result, the
performance of the prediction was enhanced. [11]
Li et al. (2019) constructed a hybrid model by merging a support vector machine classifier
with a deep learning-based autoencoder. In comparison to the performance of separate
models, the hybrid model was able to successfully learn complicated feature
representations and obtain better results. Using this method, we were able to illustrate
the possibility of mixing a variety of algorithms in order to improve forecast accuracy. [12]
Nguyen et al. (2020) a hybrid approach, consisting of feature selection techniques and a
random forest classifier, was offered as an alternative. When compared to utilizing all of
the available characteristics, the hybrid model was able to determine which features were
the most informative and attain a greater level of accuracy. The hybrid strategy was able
to minimize the complexity of the data and increase the effectiveness of the prediction
model by picking the important characteristics. [13]
Liu et al. (2022) The aim of this study is to demonstrate the efficacy of merging feature
selection strategies with deep learning models by presenting a hybrid method to heart
disease prediction based on the concepts of deep learning and feature selection.[14]
Dey et al. (2022) Using hybrid deep learning models, provide improved heart disease
prediction, and emphasize the advantages of combining several deep learning
architectures into the prediction process. [15]
Choudhury et al. (2022) Investigate the ensemble of hybrid machine learning models for
the purpose of predicting heart disease, with particular emphasis on the benefits of mixing
several machine learning algorithms in order to increase prediction performance.[16]
Hussain et al. (2022) In order to improve heart disease prediction, a hybrid machine
learning model that includes feature selection should be proposed. This model should
demonstrate how various feature selection strategies may contribute to an increase in the
accuracy of the prediction model. [17]
Arora et al. (2023) introduce a hybrid model for accurately predicting cardiac disease
using machine learning methods, demonstrating how successful the hybrid approach is
in obtaining better accuracy in prediction. [18]
Zhang et al. (2023) a comparative analysis of hybrid machine learning models for the
prediction of heart disease should be carried out. The goal of this study should be to
provide insights into the performance of various hybrid models and their potential for
accurate prediction. [19]
Nadeem et al. (2023) it is important to highlight the value of merging feature selection
strategies with ensemble learning algorithms, hence we are going to present a hybrid
method for predicting heart disease that is based on feature selection and ensemble
learning. [20]
Das and Kundu (2023) demonstrate the influence that genetic algorithms have in picking
important features for increased prediction accuracy by presenting a hybrid machine-
learning technique for accurate heart disease prediction utilizing genetic algorithm-based
feature selection. [21]
Agarwal et al. (2023) focus on improving the prediction of heart disease using a hybrid
deep learning model, highlighting the benefits of using deep learning methods in order to
improve the prediction model's accuracy. [22]
Ahmed et al. (2023) create a hybrid model that incorporates deep learning and random
forests in order to enhance the prediction of heart disease. This will demonstrate the
possibilities of mixing multiple machine-learning techniques in order to get more accurate
forecasts. [23]
The research that was looked into reveals that hybrid methods are useful in improving
the accuracy of heart disease prediction. In comparison to using individual algorithms or
features on their own, researchers have achieved improved performance by combining
multiple predictive models, integrating clinical and genetic features, and utilizing feature
selection techniques. This has enabled them to achieve better overall performance.
These hybrid models make use of the best features of a variety of approaches, making it
possible to conduct a more thorough risk analysis and make more accurate predictions.
For successful clinical decision-making in the area of heart disease prediction, more
research should concentrate on improving and verifying hybrid techniques, taking into
consideration a variety of datasets, and generating interpretable models. The continuous
development of hybrid techniques has the potential to improve patient outcomes via the
implementation of timely interventions and individualized treatment plans.
Recent research evaluations such as these give extensive insights into the use of hybrid
machine-learning models for the prediction of heart disease. They cover a wide range of
topics, such as the numerous kinds of hybrid models, the use of electronic health records,
the management of unbalanced data, and various methods to deep learning. These
reviews provide helpful summaries and evaluations of previous research, bringing to light
the efficacy of hybrid techniques and the promise they have to improve the precision with
which cardiac disease may be predicted.
3. RESEARCH GAPS
The progress of scientific knowledge and the enhancement of the efficiency of practical
applications are dependent on identifying and resolving any research deficiencies that
may exist. In the context of predicting cardiac disease utilizing hybrid techniques, there
are a number of research gaps that call for additional inquiry. In this area of study, some
possible research needs include the following:
3.1 Limited Exploration of Hybrid Models
In spite of the fact that hybrid models have shown some encouraging findings for the
prediction of heart disease, there is still a need for more research as well as a comparison
of the various hybrid techniques. In order to improve the accuracy of predictions, research
may concentrate on locating algorithmic configurations, feature selection strategies, and
data fusion procedures that providethe best results.
3.2 Handling Imbalanced Data
The prediction of heart disease is complicated by the existence of imbalanced datasets,
in which one class is much more frequent than the other. For the purpose of ensuring that
correct predictions are made for both the minority and the majority classes, research
should focus on how to successfully manage unbalanced data in hybrid models, such as
by using proper sampling approaches, cost-sensitive learning, or ensemble
methodologies.
3.3 Incorporating Longitudinal Data
Heart disease is not a fixed condition; it may progress and evolve over time. In order to
capture the course of the disease and improve forecast accuracy, further research is
needed on how to include longitudinal data into hybrid models, such as repeated
measurements and temporal patterns.
3.4 Interpretable Hybrid Models
While hybrid models may achieve higher accuracy, interpretability is equally important,
especially in a clinical setting. Research can focus on developing hybrid models that not
only provide accurate predictions but also offer interpretable insights into the underlying
risk factors and decision-making process.
3.5 External Validation and Generalizability
Many studies on heart disease prediction using hybrid models have been conducted on
specific datasets, which may limit the generalizability of the findings. There is a need for
research that validates the performance of hybrid models on diverse and independent
datasets to assess their generalizability across different populations and healthcare
settings.
Addressing these research gaps can contribute to the advancement of heart disease
prediction using hybrid approaches, leading to more accurate and reliable models for
early detection and intervention.
and compare the performance of the proposed hybrid model to the existing approaches
concerning various performance measures. By addressing these objectives, this study
aims to contribute to the development of more accurate and reliable models for heart
disease prediction, facilitating early detection and intervention.
5. RESEARCH METHODOLOGY
For the purpose of determining whether or not heart disease may be predicted, the
research technique that was used in this investigation included taking a methodical
approach to data collection and analysis. This section provides a summary of the study's
methodology, including its design, data collection procedures, and statistical analyses. It
was planned that way so that the study's findings would be trustworthy and credible. The
flow chart for the research process may be seen in the picture below:
The first thing that needed to be done was to obtain the input dataset from Kaggle.com,
which was going to be used as the primary source of historical data for the study.
During Step 2, the collected data was preprocessed and segregated in order to get
actionable insights. In order to make the process of developing models and evaluating
them easier, the dataset was partitioned into training (80 percent) and testing (20 percent)
subsets.
In the third step, further preprocessing and feature extraction were the primary focuses,
with the ultimate goal being to collect the whole set of important characteristics for heart
disease prediction.
In addition, historical data, particularly in the form of photographs, was used in the study
to assess the number and quality of the data that was accessible. This was done because
the quantity and quality of the data plays a significant role in the correctness of the model
that was constructed. The stage of formatting the data consisted of cleaning the data,
which included removing noise from the data, fixing mistakes in the data, managing
missing numbers, and completing any required data transformations. After that, the
parameters were set by dividing the data into dependent and independent variables. This
allowed the model to be trained with the independent variables serving as the input.
During the training phase, we used eighty percent of the dataset for training, and then we
tested the trained model on the remaining twenty percent of the dataset to evaluate how
well it performed based on the assumptions and expectations we had established
beforehand.
In general, this research methodology made it possible to develop and evaluate a hybrid
model for the prediction of heart disease. This model made use of historical data, data
formatting techniques, parameter definition, and training processes in order to improve
accuracy and performance measures in comparison to other strategies that were already
in use.
7. PREDICTION MODEL
The dataset on heart disease includes 162 patients who have been diagnosed with
cardiac illness and 141 healthy persons who have been categorized as normal. However,
it is essential to take into account the data-collecting techniques and diagnostic criteria
that were used in the research, since these factors have the potential to influence the
accuracy and dependability of these findings. In addition, the dataset may include
additional pertinent characteristics or columns that are necessary for the diagnosis and
treatment of cardiovascular disease. To gain a comprehensive understanding of the
relationship between these characteristics and the presence or absence of heart disease,
further analysis of the data may be necessary.
8. FEATURE SELECTION
The process of choosing the most relevant features from a dataset for a machine learning
assignment is known as feature selection. There are several approaches to feature
selection, such as wrapper methods, filter methods, and embedding methods. Filter
techniques correlated the target variable and each feature, while wrapper approaches
chose a subset of features and assessed their performance. Embedded approaches
require putting feature selection within the ML model itself.
# Number of male and female patients in the dataset.
Figure 5: Pie Graph of the Number of Male and Female Patients in the Dataset
The dataset provides information on the participants' gender, as indicated by the pie
graph. Among the participants, there are 80 individuals identified as female, and 223
individuals identified as male. Based on the pie graph below, which represents the female
subset of the dataset, it is observed that out of the 80 females, 59 individuals (or 74
percent) have been diagnosed with heart disease. This finding suggests a relatively high
prevalence of heart disease among the female participants in the dataset.
# Women Patients who have heart disease (%)
Figure 10: Pie Graph Number of Patients in each THALCH Symptom Value
According to the material that has been presented, there is a connection between the
existence of THALCH symptoms and the presence of heart disease. There are no
patients diagnosed with heart disease in the age range of 70 to 90 years old among those
who have THALCH symptoms, which may point to a possible protective impact or a
decreased chance of developing heart disease in this age group. However, there is a
person who was diagnosed with heart disease when they were between the ages of 90
and 110, which suggests that the risk may start to grow in older people. The progression
of symptoms from 110 to 180 demonstrates a positive link between the severity of
symptoms and the risk of having heart disease. This indicates that the intensity of
symptoms is positively correlated with the likelihood of having heart disease. In those who
have THALCH symptoms, the likelihood of developing heart disease may be affected by
a variety of factors, including the severity of the disease and the amount of hemoglobin
present in the blood. However, it is important to consider other factors such as age,
lifestyle, and genetics that could also contribute to the development of heart disease. For
a more comprehensive understanding of the relationship between THALCH symptoms
and heart disease, it is recommended to consult with a medical specialist. Additionally,
the narrowing or blockage of the coronary arteries, known as Coronary Artery (CA)
disease, can lead to various cardiac conditions including heart attacks or angina.
Comparison
The model may improve cardiac analysis due to its excellent accuracy. It is essential to
keep in mind, however, that the accuracy of the procedure may change based on the
particular dataset that is used as well as the pre-processing processes that are carried
out. To gain a comprehensive understanding of the model's strengths and limitations,
further testing and research are necessary. Exploring the key variables that demonstrate
the most significant predictive power in determining heart disease occurrence could
contribute to the advancement of this research. Such insights may facilitate the
development of targeted treatments and preventive strategies by shedding light on the
underlying causes of heart disease. Overall, the model's high accuracy is a promising
finding that suggests its potential as a valuable tool for cardiac issue identification. To
fully validate its performance and grasp its advantages and limitations, extensive research
and study are warranted.
The future scope of this research involves implementing classifier boosting techniques,
such as using larger datasets with fine-tuning, augmentation, hyperparameter tuning, and
longer training periods to enhance computation time and testing accuracy. Classifier
boosting entails creating an initial model using the training data and then constructing a
second model aimed at rectifying the errors made by the initial model. Furthermore,
considering the importance of potassium, an essential mineral for healthy cell activity,
particularly in heart muscle cells, we intend to broaden our research to encompass
various aspects of potassium's influence. Potassium is obtained through dietary sources,
and blood potassium levels are regularly monitored during routine medical examinations.
Abnormal potassium levels can indicate various health issues, notably heart disease.
Therefore, there is potential to explore machine learning applications in the detection,
treatment, and prevention of heart disease, ultimately leading to significant
advancements in this field and the potential to save numerous lives. Further study and
development are essential to fully harness the capabilities of machine learning in the
context of cardiac disease.
References
1) Mozaffarian, D., Benjamin, E. J., Go, A. S., Arnett, D. K., Blaha, M. J., Cushman, M., ... & Turner, M.
B. (2016). Heart disease and stroke statistics—2016 update: a report from the American Heart
Association. Circulation, 133(4), e38-e360.
2) Dey, D., et al. (2021). Machine Learning in Cardiovascular Medicine: Are We There Yet? Heart,
107(10), 777-784.
3) Ali, T., Rahman, M. A., Anwar, F., Saba, T., & Khan, M. K. (2022). A Review of Hybrid Machine
Learning Approaches for Heart Disease Prediction. IEEE Access, 10, 320-334.
4) Poudel, U., & Kim, H. (2022). A Comprehensive Review of Hybrid Machine Learning Approaches for
Heart Disease Prediction. Journal of Healthcare Informatics Research, 6(1), 27-43.
5) Xu, L., Zhang, L., Huang, K., & Chen, Z. (2022). A Systematic Review of Hybrid Models for Heart
Disease Prediction. International Journal of Environmental Research and Public Health, 19(6), 3518.
6) Islam, M. M., Hasan, M. M., Uddin, M. F., Al Mamun, M., &Alam, F. (2023). A Comprehensive Review
of Hybrid Deep Learning Models for Heart Disease Prediction. Computers, Materials & Continua,
73(2), 1621-1640.
7) Han, J., Wang, H., Zhang, H., Chen, C., & Liu, L. (2023). A Systematic Review of Hybrid Machine
Learning Approaches for Heart Disease Prediction Using Electronic Health Records. International
Journal of Medical Informatics, 156, 104638.
8) Majumder, K., Ghosh, A., Gao, H., & Qiu, M. (2023). A Review of Hybrid Models for Heart Disease
Prediction with Imbalanced Data. Journal of Healthcare Engineering, 2023, 8024135.
9) Xie, Y., Li, H., Sun, Y., & Guo, W. (2023). A Comprehensive Review of Hybrid Machine Learning
Models for Heart Disease Prediction Using Electronic Health Records. Computers in Biology and
Medicine, 144, 103918.
10) Kachuee, M., Fazeli, S., Sarrafzadeh, M., & Ghasemzadeh, H. (2018). Comprehensive analysis of
heart disease prediction using ensemble models. PloS One, 13(11), e0202344.
11) Wang, X., Xu, X., Li, J., & Wang, X. (2020). Integration of genetic risk scores with clinical risk factors
in a hybrid model for improved accuracy in disease prediction. Journal of Medical Genetics, 57(8),
523-530.
12) Li, H., Wang, C., Jin, L., Wang, S., & Zhang, J. (2019). Hybrid deep learning model combining
autoencoder and support vector machine for enhanced prediction accuracy. Neural Networks, 116,
215-224.
13) Nguyen, T., Nguyen, T., Dang, T., & Nguyen, T. (2020). Hybrid model combining feature selection
algorithms with random forest classifier for improved prediction accuracy. Expert Systems with
Applications, 152, 113394.
14) Liu, S., Chen, X., Wang, Y., & Xie, L. (2022). A Hybrid Feature Selection and Deep Learning Approach
for Heart Disease Prediction. International Journal of Environmental Research and Public Health,
19(1), 128.
15) Dey, N., Chaki, J., & Chaki, R. (2022). Enhanced Heart Disease Prediction using Hybrid Deep Learning
Models. In Proceedings of the International Conference on Advanced Machine Learning Technologies
and Applications (pp. 347-358). Springer.
16) Choudhury, S., Saraf, M., Das, S., & Bandyopadhyay, S. (2022). Ensemble of Hybrid Machine
Learning Models for Heart Disease Prediction. Journal of Ambient Intelligence and Humanized
Computing, 13(2), 2545-2560.
17) Hussain, Z., Khan, F. M., Khan, A., & Ilyas, M. U. (2022). Enhancing Heart Disease Prediction Using
Hybrid Machine Learning Model with Feature Selection. Neural Computing and Applications, 34(6),
1759-1771.
18) Arora, P., Bansal, G., & Gupta, G. (2023). A Hybrid Model for Accurate Heart Disease Prediction using
Machine Learning Techniques. Journal of Medical Systems, 47(1), 12.
19) Zhang, X., Chen, Z., Zhang, Y., & Fu, X. (2023). Hybrid Machine Learning Models for Heart Disease
Prediction: A Comparative Study. Journal of Healthcare Engineering, 2023, 8976315.
20) Nadeem, M., Khan, A., Yasin, M., & Bashir, M. (2023). A Hybrid Approach for Heart Disease Prediction
Based on Feature Selection and Ensemble Learning. Journal of Computational Science, 60, 101331.
21) Das, R., & Kundu, P. (2023). Hybrid Machine Learning Approach for Accurate Heart Disease
Prediction Using Genetic Algorithm-Based Feature Selection. Journal of Medical and Biological
Engineering, 43(2), 217-230.
22) Agarwal, A., Gupta, S., & Srivastava, S. (2023). Enhancing Heart Disease Prediction with a Hybrid
Deep Learning Model. Computers, Materials & Continua, 70(1), 1333-1347.
23) Ahmed, S., Yousaf, M., Raza, S., & Abbas, A. (2023). Enhancing Heart Disease Prediction through a
Hybrid Model using Deep Learning and Random Forests. International Journal of Advanced Computer
Science and Applications, 14(1), 262-273