Soft Voting Ensemble for Heart Disease Detection Using RF, CatBoost, and XGBoost

Title

Soft Voting Ensemble for Heart Disease Detection Using RF, CatBoost, and XGBoost

Creator

Haresh, Simon Mathew; Chalissery, Benita Jose; Newbegin, M.

Description

Global health reports constantly list heart disease as a major threat to human health. With the help of reliable information and early detection methods the disease's impact can be reduced. The study suggests an RCX(Random Forest, CatBoost, XGBoost) Ensemble machine learning method to figure out how likely a person is to get heart disease. Data for the system came through the UCI Repository containing records that were from Cleveland,Switzerland,Veterans Affairs(VA),Hungarian and statlog medical centers. As the raw datasets were not usable for performing operations, Data was put through preprocessing stages-which had involved removing missing values,making all labels uniform(standardizing them) and employing the SMOTE Technique for artificially creating new minority samples to balance the dataset. To explore an effective approach possible,models including Logistic Regression, Naive Bayes, Ran- dom Forest,CatBoost and gradient boosting(XGBoost) were trained and tested on the datasets. Using the Recursive Feature Elimination (RFE) method 8 out of 14 relevant features were selected. The models were improved through hyperparameter tuning and Among all models used the best results had come from a soft voting ensemble that combined Random Forest, CatBoost and XGBoost. The RCX ensemble model which was developed, has been shown to give accuracy and ROC AUC results of 91.18% and 96.21% respectively, showing stronger results compared to individual models. Metrics and Indicators like accuracy, precision, recall, F1-Score and the ROC AUC were used for comparing performance during testing periods. The data was visualized by using Python libraries such as matplotlib and seaborn for confusion matrices, heatmaps,and bar graphs for better understanding of the data at a glance. 2025 IEEE.

Source

Proceedings - 2025 International Conference on Transformative Computing Technologies, ICTCT 2025;pp.358-365

Date

01-01-2025

Publisher

Institute of Electrical and Electronics Engineers Inc.

Subject

Ensemble Learning; Heart Disease; Machine Learning; RCX Ensemble; SMOTE; Soft Voting

Coverage

Haresh S.M., Christ (Deemed to be) University, Department of Computer Science, Bangalore, India; Chalissery B.J., Christ (Deemed to be) University, Department of Computer Science, Bangalore, India; Newbegin M., Christ (Deemed to be) University, Department of Computer Science, Bangalore, India

Rights

Restricted Access; Hardcopy may be available in the library

Relation

ISBN: 979-833159195-3;

Format

online

Language

English

Type

Conference paper

Identifier

https://doi.org/10.1109/ICTCT69201.2025.00072

https://www.scopus.com/pages/publications/105037444599?origin=resultslist