Advancements in Heart Disease Detection:

Integrating Machine Learning Algorithms and

Clinical Insights
Sagar Sharma Ramandeep Sandhu* Manik Rakhra
School of Computer Science and School of Computer Science and School of Computer Science and
Engineering Engineering Engineering
Lovely Professional University Lovely Professional University Lovely Professional University
Phagwara, Jalandhar, Punjab, India Phagwara, Jalandhar, Punjab, India Phagwara, Jalandhar, Punjab, India
[email protected] [email protected] [email protected]

Abstract— This paper evaluates various deep learning, (EHR). These technologies provide doctors with the most
hybrid, and ensemble machine learning algorithms for heart recent information and advice, enabling them to diagnose
rhythm classification. It emphasizes how important it is to do a heart disease promptly and accurately. By putting these
comprehensive comparison analysis to ascertain the benefits cutting-edge methodologies into practice, the initiative hopes
and drawbacks of each algorithm. The study highlights the to enhance patient care related to cardiovascular health, raise
importance of feature selection in enhancing predictive diagnostic accuracy, and progress the detection of heart
capabilities, but it also draws attention to the absence of disease. Precision medicine for cardiovascular disease has
thorough evaluation metrics such as F1 score, precision, recall,
made significant progress with this.
and ROC-AUC. The study also identifies gaps in our
understanding regarding the significance of specific features in A. Machine learning Algorithms
the prognosis of heart disease and suggests more investigation
Numerous machine learning techniques, such as Naïve
into feature relevance. It is crucial to address ethical concerns
concerning data privacy, algorithmic bias, and decision- Bayes, Support Vector Machines (SVM), Random Forest,
making transparency to ensure the ethical and equitable Decision Trees, and K-nearest neighbor (KNN), are used in
implementation of predictive models in healthcare. The study the diagnosis of heart disease. These algorithms analyze
uses a broad approach that integrates ethical and technical many datasets and classify individuals according to
aspects to accelerate the development of heart illness prediction physiological, clinical, and demographic traits.
tools. The study uses a hybrid algorithm within the CDSS that
combines SVM and Random Forest with KNN to get the best
B. Electrocardiography (Ecg) Analysis
predictive model for heart disease.Keywords—Machine Several studies employ factors extracted from ECG
Learning Algorithms, Heart Disease Prediction, Feature waveforms, such as heart rate variability (HRV), changes in
Selection, Evaluation Metrics, Ethical Concerns in Healthcare. ST segments, and anomalies in rhythm, to identify cardiac
disease. Techniques like wavelet transform, discrete wavelet
I. INTRODUCTION transform, and Shannon energy envelopes are applied to
Major worldwide health issues including heart attacks process and analyze ECG data.
and strokes claim millions of lives annually and put a C. Feature Selection and Ranking
significant burden on healthcare systems. Successful
treatment of cardiac rhythm abnormalities, such as atrial Feature selection methods, including Fisher ranking,
fibrillation, bradycardia, tachycardia, and heart block, information theory, and entropy-based approaches, are
depends on prompt and accurate diagnosis. This initiative utilized to identify the most relevant features for heart
aims to combine clinical insights with machine learning disease prediction. These techniques prioritize important
algorithms to improve the efficiency and accuracy of clinical and physiological parameters, enhancing the
diagnosing heart illness. Various models are applied to accuracy and efficiency of predictive models.
evaluate various datasets containing clinical, demographic, D. Clinical Decision Support Systems (CDSS)
and physiological data. Naïve Bayes, Support Vector Some studies focus on developing clinical decision
Machines (SVM), Random Forest, Decision Trees, and K- support systems (CDSS) that integrate machine learning
nearest neighbor (KNN) are some examples of these models. algorithms with electronic health records (EHR) and patient
This research is focused on Electrocardiography (ECG) data, data. These systems assist healthcare professionals in
which are essential to comprehending heart function. The diagnosing heart disease, providing real-time insights and
research adds to a comprehensive understanding of cardiac recommendations based on patient-specific information.
illnesses and facilitates timely and accurate diagnosis by
extracting information on heart rate variability (HRV), ST II. LITERATURE SURVEY
segment shifts, and rhythm anomalies using techniques like
wavelet transform and Shannon energy envelopes. Researchers have outlined several technological
implications of using AI in the healthcare system. Below are
The research also addresses feature selection and ranking details of a few relevant works.
methods such as Fisher ranking, information theory, By
means of the predictive models are more effective when Researchers in [9] worked on detecting cardiovascular
relevant parameters for heart disease prediction are identified disease early is crucial, given its global impact. With
using entropy-based techniques. In addition, studies are being challenges in symptom identification, artificial intelligence
conducted on the development of Clinical Decision Support (AI) turns out as an essential instrument. This study suggests
Systems (CDSS), which integrate machine learning
an ensemble approach that blends the machine learning and
algorithms with patient data and electronic health records

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

deep learning models. Using Finite classification algorithms from 1991 to 2020. The analysis of 256 papers revealed an
and a publicly available dataset, the study shows the efficacy increase in the application of AI since 2008. The diagnosis of
of the RF model with a prophecy accuracy rate of 88.70%. coronary artery disease (CAD) has shown promise for
machine learning and, more recently, deep learning,
In research work [10], Cardiovascular diseases (CVDs) particularly when combined with ECG readings. Cardiac
pose significant health threats, emphasizing the importance arrhythmias, or abnormal heartbeats, pose a major risk to
of early detection to reduce mortality. The RFE-GB international health. To increase diagnosis accuracy, the
technique, or Recursive Feature Elimination-based Gradient study [16] employs the MIMIC-III database for
Boosting, is described in this article to effectively forecast comprehensive clinical monitoring data analysis. Using a
cardiac illness. With an approxy rate of 89.7% and an AUC random forest classifier, it assesses the relative importance of
of 0.84, RFE-GB performs better than other deep learning factors such as blood pressure, electrolyte levels, and
and deep learning approaches. This algorithm emerges as a respiration rate. The results confirm their significance and
promising model for CVD estimation and treatment. draw attention to their potential role in the early diagnosis of
Furthermore, the proposal discussed in the article [11] CA. Precise heart disease prediction can be achieved by
focuses on heart disease, which is a leading global cause of machine learning and data mining [17], which is noteworthy
death and for which effective treatment depends on early in regions where there is a high likelihood of misdiagnosis
detection. While machine learning (ML) has great promise, and little knowledge of cardiovascular illness. The results of
too much dimensionality and class imbalance make accurate this study show that the random forests (RF) technique
predictions difficult. This work introduces a novel approach predicts cardiac illness with 100% accuracy, sensitivity, and
that combines guided infinite feature selection and Bayesian specificity when compared to other supervised machine-
optimization with an Improved Weighted Random Forest. learning algorithms. This illustrates how basic machine
Comparative analyses demonstrating improved accuracy and learning methods can offer quite accurate diagnostic
F-measure reveal its superiority over current models. assistance. One concerning trend in the alarming rise in
Multimodal therapy is necessary for congestive heart failure heart-related mortality is modern lifestyles. The research
(CHF), and it might be particularly challenging for older conducted in [18] is predicated on the early detection of
individuals who also co-occur anxiety and depression [12]. cardiac disease, which is a major global cause of death.
The widespread functionalities and verification of the CHF Machine learning, in particular the Random Forest method,
self-management smartphone apps now available are proved important for accurate illness identification. This
lacking. The HeartMan project skillfully combines artificial approach, which emphasizes excellent performance,
intelligence and patient monitoring to provide exceptional flexibility, and success rates, focuses on predicting cardiac
accuracy. problems early on using prior patient records. The
A proof-of-concept trial with 56 patients reveals significant researchers in [19] focus on cardiovascular diseases (CVDs),
improvements in self-care behavior and psychological well- which claim millions of lives annually and emphasise the
being, showcasing the efficacy of the HeartMan system. importance of early detection and surveillance. Even with the
Moreover, the work done in [13] introduces a resource- growing number of biometric signal recording devices,
efficient system for extracting crucial features from analyzing electrocardiogram (ECG) data by hand still takes a
electrocardiogram (ECG) signals. Utilizing two delta-sigma long time. This study looks at the advantages and
modulators and three wave detection algorithms, the system disadvantages of the technologies that are already in use and
encodes essential heartbeat details into 68 bits, a mere offers suggestions for further research aimed at developing
1.48% of comparable methods. Employing patient-specific automatic anomaly detection in ambulatory monitoring
training and the MIT-BIH database, random forest systems for precise heart condition diagnosis. In [20], a
classifiers achieve high F1 scores of 81.05% for brand-new technique for identifying coronary artery disease
supraventricular ectopic beats (SVEB) and 97.07% for (CAD) utilizing heart rate variability (HRV) data is
ventricular ectopic beats (VEB). The method, with its time presented. The paper adds multiscale wavelet packet
efficiency and low complexity, offers a reliable solution for transform and entropy features to extreme learning machines
ECG signal analysis, particularly applicable to wearable and (ELM), generalized discriminant analysis (GDA), and Fisher
implanted medical devices for arrhythmia classification. ranking for accurate CAD recognition. The experimental
Artificial intelligence (AI) is revolutionizing clinical care, results show superior performance; the proposed model
drug research, diagnosis, therapy, and risk prediction in achieves 100% accuracy by using features with the highest
cardiovascular medicine, according to an article [14]. AI rank in entropy. According to a study published in [21], heart
encompasses machine learning and deep learning. Early failure is a prevalent clinical issue that needs to be identified
identification is critical since heart failure has a high death early for the best possible management. An
and prevalence rate. Neural network models can achieve electrocardiogram (ECG) is one frequently used diagnostic
85% accuracy in this regard. Artificial intelligence (AI) technique. This study examines several machine learning
supports the processing of data from cardiac imaging and techniques for diagnosing cardiac disease, including
electrocardiograms by using algorithms for decision-making enhanced SVM, DBSCAN, and SMOTE-ENN-using clinical
models. The optimization of cardiovascular outcomes has decision support system, Naïve Bayes, and support vector
been achieved by utilizing decision trees, logistic regression, machines using XGBoost . The project employs machine
and rough sets in the diagnosis, prevention, and therapy of learning to enhance clinical decision-making and patient
heart failure. The gold standard, coronary angiography, was outcomes, with a focus on early detection. Based on coronary
applied in [15]. offers risks related to coronary artery disease artery disease (CAD), a global health concern, the study by
(CAD). Artificial intelligence (AI) offers a non-intrusive [22] emphasizes the significance of an appropriate diagnosis
method called test probability. This study uses artificial and treatment plan. This article includes a thorough
intelligence (AI) to analyze global trends in CAD detection evaluation of the literature on artificial intelligence
classification algorithms for CAD detection. The efficacy of G. Clinical Decision Support System (CDSS)
machine learning methods such random forests, K-nearest  Utilize clinical data, such as electronic health
networks, neural networks, support vector machines, and records (EHR) and patient demographics, in
linear regression is highlighted in the review, which combination with the KNN predictions to provide
comprised 54 publications. These models demonstrate how decision support.
CAD diagnosis could be carried out more quickly and
 Develop rules or algorithms based on medical
accurately, cutting down on mistakes, diagnostic time, and
guidelines to interpret KNN predictions and provide
related expenses. The work proposed in [23] is based on the
recommendations for diagnosis and treatment.
classification and regression tree (CART) technique in
supervised machine learning to forecast cardiac illness and H. Model Evaluation
offer decision rules. This makes it possible to establish  Examine the efficacy of the KNN model with the
transparent links between input and output variables. The test data.
model shows its reliability with an accuracy rate of 87% and
offers simplified clinical applications.  Utilise assessment criteria such as recall, accuracy,
By addressing diagnosis-related concerns, this approach precision, and F1-score.
assists patients and healthcare professionals in managing I. Clinical Decision Support Evaluation
heart disease within time and financial constraints. [24]–[25].  Assess the effectiveness of the CDSS in providing
III. RESEARCH METHODOLOGY decision support for heart illness prognosis.
Various phases of the proposed approach are listed  Assess the precision of CDSS guidelines. compared
below: to ground truth diagnoses or expert opinions.
A. Data preprocessing J. Comparison and Selection
 Load the heart disease dataset.  Compare the performance of the SVM and
ensemble models.
 Before encoding categorical variables, address
missing values and preprocess the data.  Select the model with the best performance based
on the chosen evaluation metric(s).
 Split the dataset into features (X) and target variables
(y).
Table 1: Comparison analysis of various algorithms
B. Feature Scaling
 Standardize or normalize the features to bring them to Algorithm K-Nearest Hybrid Support
a similar scale. name Neighbors Vector Machine
 This step is crucial for SVM to perform effectively. (KNN) (SVM) and Random
Parameter Forest (RF)
C. Train/Test Split used Ensemble
During this phase, To assess the effectiveness of the
models, divide the dataset into training and testing sets. Training Requires labeled Requires labeled data
Data data
Interpretab Fairly SVM may be less
D. SVM Model ility interpretable due interpretable, Random
to simple Forest less so
 Initialize a Support Vector Machine classifier. algorithm
 Choose appropriate hyperparameters such as the Performan Performance SVM and Random
kernel type, regularization parameter, etc. ce depends on the Forest generally
 Train the SVM model on the training data. quality of features perform well,
and data, and may especially with
not generalize optimized
E. Ensemble Learning well hyperparameters
Scalability Can be SVM and Random
 Initialize an ensemble classifier (e.g., Random computationally Forest can handle
Forest).
expensive for larger datasets
 Choose appropriate hyperparameters such as the large datasets due efficiently
number of estimators, maximum depth, etc. to the need to
 Train the ensemble model on the training data. calculate
distances for each
 Use the ensemble model's training set of data. prediction
F. KNN Model Sensitivity Sensitive to noisy SVM and Random
to Noise data Forest can handle
 Initialize a K-Nearest Neighbors classifier. noise to some extent,
 Choose the number of neighbors (K) and other especially Random
hyperparameters. Forest
Ease of Fairly easy to SVM and Random
 Train the KNN model on the training data.
Implement implement Forest may require
ation more parameter
 tuning and Table 2: Accuracy parameter values using various health
optimization parameters
Use in Can provide May require
Clinical straightforward additional
Setting recommendations interpretation and
based on patient validation due to the
data complexity
Overall Suitable for Suitable for more
simpler datasets complex datasets and
and quick potentially more
decision-making, accurate predictions,
but may lack especially with
generalization optimization

Table 1 provides various important features based on the

parameters enlisted in column 1. Classification model predicting various heart conditions.
Accuracy, precision, recall (sensitivity), F1-score, and
support are explained. A table interprets these metrics for
IV. RESULTS AND GRAPHS
different heart conditions: Angina, Atrial Fibrillation,
First of all, a primary dataset is used for the Coronary Artery Disease, Healthy, Heart Failure, and
implementation of the proposed model. This dataset was Myocarditis. Each class's strengths and weaknesses are
collected from highlighted through an analysis of precision, recall, and F1-
https://data.cdc.gov/api/views/i2vk-mgdh/rows.csv? score. The total effectiveness of the model is shown by the
accessType=DOWNLOAD. Figure 1 gives an overview of weighted average precision, recall, and F1 score for each
the collected dataset. class.

In summary, the model demonstrates high accuracy and

precision for most classes but lower recall for some,
suggesting room for improvement in identifying certain
conditions.

Figure 1: Overview of input dataset to train the proposed

model
The information acquired displays patient data that is
utilized by a machine learning-based Clinical Decision
Support System (CDSS) to predict heart diseases. It includes
metrics such as age (42-65 years), gender, cholesterol level
(180-300 mg/dL), blood pressure (115/70 to 155/95 mmHg),
blood sugar level (105-150 mg/dL), and family history of
heart disease. The machine learning predictions, either
Positive or Negative, are compared against clinical
diagnoses like Myocardial infarction, heart failure, ischemic
heart disease (IHD), and overall health, Angina,
Myocarditis, and Atrial Fibrillation. Positive predictions Figure 2: Visual representation of the prevalence of
often correlate with higher health metrics and family history, various heart conditions within a sample population
leading to serious diagnoses, while Negative predictions
typically indicate healthier profiles. The data is utilized to A visual depiction of the prevalence of different cardiac
problems within a sample population is provided by the pie
support early detection and treatment and improve the chart in Figure 2, which examines the clinical diagnosis
accuracy of heart disease prediction using K-Nearest distribution. Consider the following while interpreting the
Neighbours (KNN) and a hybrid model that combines chart in order to enhance algorithms for the identification of
heart disease:
Random Forest and Support Vector Machine (SVM).
The proposed approach is a hybrid approach using KNN
and SVM algorithms. Table 2 as shown below provides the
values of four accuracy parameters: precision, recall, f1-
score, and support. These values are generated based on
different health parameters.
  Healthy (33.1%): This is the highest number,  Zero Values: Cells with zero values show that
indicating that one-third of the sample's individuals there were no instances of a particular
are in excellent health. This highlights the need for misclassification, which is ideal.
a model that accurately assesses the absence of  Overall Accuracy: By dividing the total number of
heart illness in order to avoid false positive results. predictions by the sum of the diagonal values, the
 Angina (15.0%) and Coronary Artery Disease overall accuracy of the model is computed.
(17.5%): These conditions should be well-
represented in the training set due to their relative Figure Figure 4 shows the Support Vector Machine
prevalence. The models should be able to (SVM) Confusion Matrix, an essential diagnostic tool
for evaluating a diagnostic system's functionality. To
distinguish between different circumstances build a reliable diagnostic system, it is necessary to
according to their sensitivity. determine the model's performance in many scenarios,
and this matrix is crucial for doing so. It assists in
 Heart failure (9.5%), myocarditis (8.5%), determining the benefits of the model as well as
myocardial infarction (8.0%), and atrial potential areas for development, such as increasing
fibrillation (8.0%): These less frequent sensitivity for particular circumstances or lowering the
occurrences suggest that the model needs to possibility of misdiagnosing healthy people.
adequately account for class imbalance. To ensure
that specific situations are not underrepresented
throughout the model's training phase, techniques
like oversampling, undersampling, or synthetic data
generation might be used.
 High Dimensionality: Because of the variety of
scenarios, feature selection becomes crucial to
improve computing performance and prevent
overfitting. The main focus of the model should be
on the qualities that best describe each condition.

As shown in Figure 3, an effective tool for assessing a

machine learning-based diagnostic system is the Random
Forest Confusion Matrix. Below is an explanation of Figure
3:

 True Positives (Diagonal): The diagonal values

show which predictions made by the model were
correct. For example, the value of "16" for
"Healthy" means that the model accurately Figure 4: A thorough examination of the predicted
identified 16 individuals as healthy. outcomes based on the SVM Confusion Matrix
Figure 4, which highlights the development of
dependable diagnostic systems that are practical in real-
world settings, is explained as follows:
 True Positives (Diagonal): The scenarios where
the model's predictions were realized are shown by
the values on the diagonal, which go from "4" for
angina to "16" for healthy.
 False Positives and False Negatives (Off-
Diagonal): Numbers off the diagonal, like the ‘2’
for Healthy predicted as Atrial Fibrillation, show
where the model made errors. These are crucial for
understanding where the model might confuse one
condition for another.
 Zero Values: Cells with zero values, especially
outside the diagonal, are ideal as they indicate no
misclassifications between those particular
conditions.
 Class Imbalance: The varying numbers across the
Figure 3: A thorough examination of the Random Forest matrix suggest a class imbalance that the model
Confusion Matrix forecasts' outcomes must handle to avoid bias towards more frequent
classes.
 False Positives and False Negatives (Off-
Diagonal): Numerical forecasts that are not correct For developing diagnostic systems that aid in early
are displayed diagonally. For instance, the number diagnosis and enhance clinical decision-making:
"1" in the "Healthy" row and "Atrial Fibrillation"
column denotes the misdiagnosis of atrial
fibrillation in one healthy individual.
  Accuracy: Focus on increasing the diagonal Ranking and TBW Scheduling Method," in International
Conference on Wireless Sensor Networks, Ubiquitous
values, which represent correct predictions, to Computing and Applications, vol. 244, 2022. doi:
improve the overall accuracy. 10.1007/978-981-16-2919-8.
 Precision and Recall: Work on reducing false [7] Kaur, M., Aron, R., Wadhwa, H., Tandon, R., Oo, H. N., &
Sandhu, R. (2024). A Comprehensive Exploration of
positives and false negatives to improve these Machine Learning and IoT Applications for Transforming
metrics, which are vital for a reliable diagnostic Water Management. Innovations in Machine Learning and
system. IoT for Water Management (pp. 25-50). IGI Global.
https://doi.org/10.4018/979-8-3693-1194-3.ch002.
 Feature Selection: Analyze the misclassifications [8] K. Lakhwani, G. Sharma, R. Sandhu, N. K. Nagwani, S.
to identify if additional features or better feature Bhargava, V. Arya, and A. Almomani, "Adaptive and
selection could improve the model’s predictive Convex Optimization-Inspired Workflow Scheduling for
power. Cloud Environment," International Journal of Cloud
Applications and Computing (IJCAC), vol. 13, no. 1, pp. 1-
 Clinical Integration: Ensure the model’s insights 25, 2023. doi: 10.4018/IJCAC.324809.
are interpretable by clinicians to make the system [9] A. Alqahtani, S. Alsubai, M. Sha, L. Vilcekova, and T.
Javed, “Cardiovascular Disease Detection using Ensemble
practical and trustworthy. Learning,” Comput Intell Neurosci, vol. 2022, 2022, doi:
10.1155/2022/5267498.
The aforementioned concerns can be addressed, and the [10] P. Theerthagiri, “Cardiovascular Disease Prediction using
diagnostic framework will be enhanced to facilitate Recursive Feature Elimination and Gradient Boosting
clinical judgment and prompt diagnosis, hence Classification Techniques.”
enhancing patient results and facilitating more tailored [11] A. Abdellatif, H. Abdellatef, J. Kanesan, C. O. Chow, J. H.
treatment plans. Chuah, and H. M. Gheni, “Improving the Heart Disease
Detection and Patients’ Survival Using Supervised Infinite
Feature Selection and Improved Weighted Random Forest,”
IEEE Access, vol. 10, pp. 67363–67372, 2022, doi:
Conclusion 10.1109/ACCESS.2022.3185129.
In general, this study highlights the noteworthy [12] A. Baert et al., “A Personal Decision Support System for
advancements in the field of cardiac illness identification by Heart Failure Management (HeartMan): Study protocol of
incorporating machine learning algorithms with clinical the HeartMan randomized controlled trial,” BMC
observations. Through the utilization of several models, this Cardiovasc Disord, vol. 18, no. 1, Sep. 2018, doi:
10.1186/s12872-018-0921-2.
work shows potential to improve diagnostic accuracy and
precision by using ensemble methods, Naïve Bayes, Support [13] B. H. Kung, P. Y. Hu, C. C. Huang, C. C. Lee, C. Y. Yao,
and C. H. Kuan, “An Efficient ECG Classification System
Vector Machines (SVM), Random Forest, and better feature Using Resource-Saving Architecture and Random Forest,”
selection procedures. Further enhancing the diagnostic IEEE J Biomed Health Inform, vol. 25, no. 6, pp. 1904–
capabilities are the analysis of Electrocardiography (ECG) 1914, Jun. 2021, doi: 10.1109/JBHI.2020.3035191.
signals and the creation of Clinical Decision Support [14] F. Yasmin et al., “Artificial intelligence in the diagnosis and
Systems (CDSS), which allow for rapid intervention and detection of heart failure: the past, present, and future,”
Reviews in Cardiovascular Medicine, vol. 22, no. 4. IMR
individualized patient treatment. Nonetheless, issues like the Press Limited, pp. 1095–1113, Dec. 22, 2021. doi:
requirement for thorough assessment metrics and the 10.31083/j.rcm2204121.
resolution of moral questions about algorithmic bias and [15] R. Alizadehsani et al., “Coronary artery disease detection
data privacy continue to exist. Going forward, to guarantee using artificial intelligence techniques: A survey of trends,
the proper use of technology in cardiovascular healthcare, geographical differences and diagnostic features 1991–
2020,” Computers in Biology and Medicine, vol. 128.
research endeavors should concentrate on enhancing Elsevier Ltd, Jan. 01, 2021. doi:
interpretability, strengthening predictive models, and 10.1016/j.compbiomed.2020.104095.
addressing ethical issues. By working together, we can move [16] S. S. Yadav and S. M. Jadhav, “Detection of common risk
closer to a time when heart disease will be far less common, factors for diagnosis of cardiac arrhythmia using machine
and patient outcomes will be better due to early detection learning algorithm,” Expert Syst Appl, vol. 163, Jan. 2021,
and care. doi: 10.1016/j.eswa.2020.113807.
[17] M. M. Ali, B. K. Paul, K. Ahmed, F. M. Bui, J. M. W.
References Quinn, and M. A. Moni, “Heart disease prediction using
supervised machine learning algorithms: Performance
[1] Smith, et al., "Prediction of Cardiovascular Disease Using
analysis and comparison,” Comput Biol Med, vol. 136, Sep.
Machine Learning Techniques: A Review," Journal of 2021, doi: 10.1016/j.compbiomed.2021.104672.
Cardiology, vol. 15, no. 3, pp. 210-225, 2023.
[2] J. Doe, et al., "Application of Machine Learning Algorithms [18] M. S. Raja, M. Anurag, C. P. Reddy, and N. R. Sirisala,
in Identifying Risk Factors for Coronary Artery Disease," “Machine Learning Based Heart Disease Prediction
Cardiovascular Research, vol. 28, no. 2, pp. 115-130, 2023. System,” in 2021 International Conference on Computer
[3] M. Faiz, R. Sandhu, M. Akbar, A. A. Shaikh, C. Bhasin, Communication and Informatics, ICI 2021, Institute of
and N. Fatima, "Machine Learning Techniques in Wireless Electrical and Electronics Engineers Inc., Jan. 2021. doi:
Sensor Networks: Algorithms, Strategies, and 10.1109/ICCCI50826.2021.9402653.
Applications," International Journal of Intelligent Systems [19] H. Zu Li and P. Boulanger, “A survey of heart anomaly
and Applications in Engineering, vol. 11, no. 9s, pp. 685- detection using ambulatory electrocardiogram (ECG),”
694, 2023. Sensors (Switzerland), vol. 20, no. 5. MDPI AG, Mar. 01,
[4] D. Johnson, et al., "Machine Learning Approaches for Early 2020. doi: 10.3390/s20051461.
Detection of Heart Failure: A Comparative Study," Journal
of Cardiac Imaging, vol. 7, no. 4, pp. 320-335, 2023. [20] R. S. Singh, B. S. Saini, and R. K. Sunkaria, “Detection of
[5] E. Brown, et al., "Predicting Stroke Risk Using Machine coronary artery disease by reduced features and extreme
Learning Models: A Systematic Review and Meta- learning machine,” Clujul Medical, vol. 91, no. 2, pp. 166–
Analysis," Stroke Research and Treatment, vol. 12, no. 1, 175, 2018, doi: 10.15386/cjmed-882.
pp. 45-60, 2023. [21] U. Nagavelli, D. Samanta, and P. Chakraborty, “Machine
[6] R. Sandhu and K. Lakhwani, "Improved Scientific Learning Technology-Based Heart Disease Detection
Workflow Scheduling Algorithm with Distributed Heft
 Models,” J Healthc Eng, vol. 2022, 2022, doi:
10.1155/2022/7351061.
[22] A. Garavand, A. Behmanesh, N. Aslani, H. Sadeghsalehi,
and M. Ghaderzadeh, “Towards Diagnostic Aided Systems
in Coronary Artery Disease Detection: A Comprehensive
Multiview Survey of the State of the Art,” International
Journal of Intelligent Systems, vol. 2023. Wiley-Hindawi,
2023. doi: 10.1155/2023/6442756.
[23] R. Sandhu, H. Kaur, H. S. Sohal, K. Handa, A. Singh, and
S. M. N. Islam, "Enhancement in Cloud Performance using
the Clustering Method of Scientific Workflow Tasks," in
2023 2nd International Conference on Ambient Intelligence
in Health Care (ICAIHC), pp. 1-6, 2023.
[24] M. Ozcan and S. Peker, “A classification and regression tree
algorithm for heart disease modeling and prediction,”
Healthcare Analytics, vol. 3, Nov. 2023, doi:
10.1016/j.health.2022.100130.
[25] R. Sandhu, K. Lakhwani, “Enhanced Scientific Workflow
Scheduling in Cloud System,” International Conference on
Communications and Cyber-Physical Engineering (ICCCE
21). Vol: 828, 2021, pp: 133-139.
https://doi.org/10.1007/978-981-16-7985-8_14.