1Department of Computer Science & Engineering, Anjuman College of Engineering & Technology, Nagpur, India. Email: ansari.nargisrana@gmail.com
2Department of Computer Science & Engineering, Anjuman College of Engineering & Technology, Nagpur, India. Email: arazzak@anjumanengg.edu.in
Background: Coronary heart disease (CHD) is a multifactorial issue with early prediction being a critical problem to address because of its high levels of imbalance in classes when using a large scale of healthcare data. This paper is a complete machine learning and deep learning model of CHD risk prediction based on population-level data on health records. An analysis using numerous exploratory features was done on a dataset of 246,013 records with demographic, clinical, lifestyle and behavior attributes.
Methods: To overcome the biased ratio of the CHD-positive and the CHD-negative data, Synthetic Minority Oversampling Technique (SMOTE) was introduced following the dataset division stage, leading to the balanced distribution of the training material. There are several classification models created such as Decision Tree (DT), Random Forests (RF), AdaBoost, and Voting Ensemble classifier which are compared with a proposed sequential neural network. The dropout regularization and the Adam optimizer were utilized to improve the generalization and convergence stability of the neural network architecture. Accuracy, precision, recall, F1 score, confusion matrices, and ROC-AUC curves were used to measure the quality of the models.
Results: The findings of the experimental prove that the proposed sequential neural network with SMOTE is superior to all the baseline models, with the accuracy and precision of 95.96, recall of 96.17, and F1 score of 96.16.
Conclusion: The investigation outcomes prove that integration of imbalance-conscious learning and deep representation modeling provide a high degree of predictivity of CHD, with a valid and scalable abiding tool of the decision-support consideration of the initial cardiovascular risk assessment.
Keywords: Coronary Heart Disease, Machine Learning, Healthcare Analytics, Risk Prediction, SMOTE
How to cite this article: Ansari NRAW, Razzaque A. Next-Generation Healthcare Analytics for Imbalance-Aware Early Coronary Heart Disease Prediction Using SMOTE-Enhanced Model. Int J Drug Deliv Technol. 2026;16(13s): 1043-1044. DOI: 10.25258/ijddt.16.13s.114.
Source of support: Nil.
Conflict of interest: None