|
|
 |
| |
|
|
|
Detecting Heart Attacks Using Learning Classifiers |
|
|
|
PP: 2859-2875 |
|
|
|
doi:10.18576/isl/120715
|
|
|
|
Author(s) |
|
|
|
Manar M. Hafez,
Ramy S. Ebeid,
Diana T. Mosa,
|
|
|
|
Abstract |
|
|
| Cardiovascular diseases (CVDs) have emerged as a critical global threat to human life. The diagnosis of these diseases presents a complex challenge, particularly for inexperienced doctors, as their symptoms can be mistaken for signs of aging or similar conditions. Early detection of heart disease can help prevent heart failure, making it crucial to develop effective diagnostic techniques. Machine Learning (ML) techniques have gained popularity among researchers for identifying new patients based on past data. While various forecasting techniques have been applied to different medical datasets, accurate detection of heart attacks in a timely manner remains elusive. This article presents a comprehensive comparative analysis of various ML techniques, including Decision Tree, Support Vector Machines, Random Forest, Extreme Gradient Boosting (XGBoost), Adaptive Boosting, Multilayer Perceptron, Gradient Boosting, K-Nearest Neighbor, and Logistic Regression. These classifiers are implemented and evaluated in Python using data from over 300 patients obtained from the Kaggle cardiovascular repository in CSV format. The classifiers categorize patients into two groups: those with a heart attack and those without. Performance evaluation metrics such as recall, precision, accuracy, and the F1-measure are employed to assess the classifiers’ effectiveness. The results of this study highlight XGBoost classifier as a promising tool in the medical domain for accurate diagnosis, demonstrating the highest predictive accuracy (95.082%) with a calculation time of (0.07995 sec) on the dataset compared to other classifiers.
|
|
|
|
|
|
|
|
