Wrapper method for feature selection to classify cardiac arrhythmia.

Area of Science:

• Biomedical Engineering
• Machine Learning in Healthcare
• Cardiology

Background:

• Efficient monitoring of cardiac patients is crucial for saving lives.
• Cardiac disease prediction and classification are increasingly significant.
• Arrhythmia classification is a key area for improving cardiac patient care.

Purpose of the Study:

• To develop and evaluate a predictive model for classifying cardiac arrhythmias.
• To compare the performance of various machine learning classifiers for arrhythmia detection.

Main Methods:

• Feature selection using a wrapper algorithm around Random Forest.
• Implementation and cross-validation of Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Naïve Bayes, Random Forest, and Multi-Layer Perceptron (MLP).
• Utilized the Cardiac Arrhythmia dataset from the UCI machine learning repository with 10-fold cross-validation.

Main Results:

• Multi-Layer Perceptron (MLP) achieved the highest average accuracy of 78.26%.
• K-Nearest Neighbor (KNN) and Support Vector Machine (SVM) showed accuracies of 76.6% and 74.4%, respectively.
• The proposed MLP model demonstrated superior performance compared to previous models.

Conclusions:

• The developed predictive model, particularly MLP, shows strong potential for accurate cardiac arrhythmia classification.
• Machine learning approaches can significantly enhance the efficiency of cardiac patient monitoring and diagnosis.
• This study contributes to advancing automated methods for detecting and classifying heart rhythm disorders.