Hybrid CNN-Clinical Data Fusion with Cost-Sensitive Learning for Robust and Fair Heart Disease Detection

C Raja Sekhar; B Thasmiya Sulthana; B Eswar Naik; M Deepthi; K Sudharsan Reddy

doi:10.70528/IJLRP.v7.i3.2010

Hybrid CNN-Clinical Data Fusion with Cost-Sensitive Learning for Robust and Fair Heart Disease Detection

Author(s)	C Raja Sekhar, B Thasmiya Sulthana, B Eswar Naik, M Deepthi, K Sudharsan Reddy
Country	India
Abstract	Early detection of heart disease is crucial for improving patient outcomes and reducing global healthcare burdens. The existing system employs EfficientNet-B3, a Computer Neural Networks (CNN) architecture, to analyze retinal fundus images for predicting heart disease risk. It combines retinal images with clinical data (blood pressure, cholesterol, Body Mass Index (BMI), diabetes) through multimodal fusion. The model uses global average pooling, dropout, and batch normalization to improve generalization and limit overfitting. Data augmentation methods like adaptive histogram equalization and gamma correction enhance feature extraction and robustness. The system attains high accuracy and Area Under the Curve - Receiver Operating Characteristic (AUC-ROC) (above 96%), but depends on large labeled datasets and requires significant computational power. Interpretability is provided by Gradient-weighted Class Activation Mapping (Grad-CAM) and SHapley Additive exPlanations (SHAP), yet challenges remain with false negatives and generalization across diverse populations. The proposed system focuses on enhancing model interpretability and generalizability by integrating attention-based Computer Neural Networks (CNN) architectures and hybrid feature integration. It aims to use a more diverse dataset to minimize biases and improve performance across various populations. The system will also explore multimodal data fusion, combining retinal imaging with additional health parameters to further boost accuracy. Cost-sensitive learning and ensemble techniques will be employed to reduce false negatives and improve sensitivity. The model will be validated across multiple clinical settings to ensure robustness and fairness.
Keywords	EfficientNet-B3, Cost-sensitive learning, Ensemble techniques, Generalization, Multimodal data fusion.
Field	Engineering
Published In	Volume 7, Issue 3, March 2026
Published On	2026-03-20
DOI	https://doi.org/10.70528/IJLRP.v7.i3.2010
Short DOI	https://doi.org/hbtbqw

View / Download PDF File

doi

CrossRef DOI is assigned to each research paper published in our journal.

IJLRP DOI prefix is
10.70528/IJLRP

Downloads

Research Paper Format Copyright Permission Form and Undertaking Form Cover Page Vol 6 Isu 12 Cover Page Vol 6 Isu 10 Cover Page Vol 6 Isu 9

All research papers published on this website are licensed under Creative Commons Attribution-ShareAlike 4.0 International License, and all rights belong to their respective authors/researchers.

CC-BY-SA

About IJLRP Fees & Payment Current Issue Publication Archive	Submit Research Paper Track Submission Status Publication Guidelines Publication Ethics Peer Review & Plagiarism	Join as a Reviewer Editors & Reviewers Reviewer Referral Program Get Reviewer Membership Certi.	Website/Journal Policies Usage Policy Content Policies Privacy Policy

Contact Us		+91-9687-828-838	editor@ijlrp.com

International Journal of Leading Research Publication

A Widely Indexed Open Access Peer Reviewed Multidisciplinary Monthly Scholarly International Journal

Hybrid CNN-Clinical Data Fusion with Cost-Sensitive Learning for Robust and Fair Heart Disease Detection

Share this