Early Prediction of Heart Disease Using a Stacking Ensemble Machine Learning Model
Students & Supervisors
Student Authors
Supervisors
Abstract
Cardiovascular disorders contribute heavily to worldwide mortality rates; the development of accurate and robust preventive screening systems has become increasingly essential. In this paper, we propose a two-tier stacking ensemble architecture that predicts heart disease using structured clinical data. The proposed architecture consists of RF and XGBoost, which are selected as diverse base learners to identify complex nonlinear patterns and high-order interactions when modeling medical datasets. Rather than using the individual model predictions, in this pipeline, the prediction probabilities from both foundational classifiers are combined and treated as meta-features. In the second stage, a logistic regression meta-classifier is trained on these probability-level representations to find an optimal decision boundary that can preserve the complementary benefits of both tree-based ensembles and linear classifiers. This stochastic ensemble method enhances generalization capability at the same time with mitigating overfitting and diversifying the independence of individual models. The final categorization is done using a calibrated decision threshold on the probability scores at the meta level, which enables optimization for sensitivity in screening-focused clinical scenarios. Results show that probability-based stacking ensembles perform as a highly scalable, interpretable, and accurate solution for early diagnosis of heart disease with great potential to be successfully integrated into progressive clinical decision support systems.
Keywords
Publication Details
- Type of Publication:
- Conference Name: International Conference on Electrical, Computer and Communication Technologies (ECCT 2026)
- Date of Conference: 05/07/2026 - 05/07/2026
- Venue: Dhaka International University, Bangladesh
- Organizer: Dhaka International University, Bangladesh