Weight Initialization-Based Rectified Linear Algorithm for Accurate Prediction of Chronic Heart Disease Compared with PCHF Feature Engineering Technique

S. S. Deepak Senni, M. Krishnaraj

2025

Abstract

Cardiovascular disease continues to pose a significant challenge to global health, underscoring the critical need for early and precise prediction to enable effective preventive strategies. This paper investigates the promising role of supervised learning techniques within the realm of Artificial Intelligence (AI) for predicting heart disease. We explore notable advancements in various algorithms, including logistic regression (LR), support vector machines (SVM), and deep neural networks (DNN), emphasizing their ability to uncover intricate patterns within extensive medical datasets. Moreover, the research highlights the capacity of AI-enhanced cardiologists to analyze a wide array of patient data, encompassing demographics, medical histories, lab test outcomes, and ECG readings. Such comprehensive evaluations promise to enhance the accuracy and personalization of risk assessments, potentially facilitating earlier interventions and improving patient outcomes. This study also addresses the significant challenges related to data quality, the mitigation of biases, and the explainability of AI models, highlighting the need for ethical considerations in their design and deployment. We classify ECG stages utilizing two models: a Cardiology model based on Machine Learning techniques with a specific dataset and a Deep Learning Model focused on identifying cardiovascular disease through ECG image classification. Additionally, the application of the Weight Initialization-Based Rectified Linear Algorithm (WiReL) for heart disease prediction underscores the integration of optimized weight initialization principles along with ReLU activation within a deep learning context. Our findings demonstrate that the WiReL algorithm outperforms the Principal Component Heart Failure (PCHF) Feature Engineering Technique in terms of predictive accuracy. Furthermore, this paper discusses potential future advancements in AI-driven heart disease prediction, considering the implications of emerging methodologies such as Generative AI and federated learning to further enhance this vital field. Our proposed research offers meaningful contributions to medical science and its endeavors in combating cardiovascular disease.

Paper Citation

in Harvard Style

Senni S. and Krishnaraj M. (2025). Weight Initialization-Based Rectified Linear Algorithm for Accurate Prediction of Chronic Heart Disease Compared with PCHF Feature Engineering Technique. In Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25; ISBN 978-989-758-777-1, SciTePress, pages 618-626. DOI: 10.5220/0013902900004919

in Bibtex Style

@conference{icrdicct`2525,
author={S. Senni and M. Krishnaraj},
title={Weight Initialization-Based Rectified Linear Algorithm for Accurate Prediction of Chronic Heart Disease Compared with PCHF Feature Engineering Technique},
booktitle={Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25},
year={2025},
pages={618-626},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013902900004919},
isbn={978-989-758-777-1},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25
TI - Weight Initialization-Based Rectified Linear Algorithm for Accurate Prediction of Chronic Heart Disease Compared with PCHF Feature Engineering Technique
SN - 978-989-758-777-1
AU - Senni S.
AU - Krishnaraj M.
PY - 2025
SP - 618
EP - 626
DO - 10.5220/0013902900004919
PB - SciTePress