Multi-Modal AI Framework for Cardiovascular Risk Prediction

View at publisher Summarise with VIRAT
Authors: G. Dinesh, P. Jaidev, B. R. Devaj, S. Nagendra Prabhu
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

One of the leading causes of morbidity and death worldwide is still cardiovascular disease (CVD), which emphasizes the need for fast, accurate, and noninvasive diagnosis techniques. In this study, we propose a hybrid deep learning framework that integrates retinal image analysis with patient-specific clinical data to improve CVD classification. Our approach employs two pre-trained convolutional neural networks, VGG16 and AlexNet, to extract salient features from retinal fundus images, capturing both hierarchical and spatial patterns indicative of cardiovascular risk. These image-derived features are subsequently fused with clinical parameters—such as blood pressure, cholesterol levels, glucose measurements, and lifestyle factors—to form a comprehensive multimodal representation of the patient’s health status. After extensive training and validation, the model is deployed via the Streamlit-based web application, enabling clinicians and researchers to upload retinal images and input clinical data for real-time CVD risk assessment. Experimental results demonstrate high classification accuracy, with robust F1-scores and a significant area beneath the ROC-AUC receiver operational characteristic curve. The minimal gap between training and validation metrics confirms effective generalization. Overall, our hybrid system highlights the promise of combining noninvasive retinal imaging with structured clinical data to enhance early CVD detection, thereby improving clinical decision-making and patient outcomes.

Fields of Study & Subjects

Artificial Intelligence in Healthcare Retinal Imaging and Analysis Health Promotion and Cardiovascular Prevention

Publication details

Journal / source
CRC Press eBooks
Pages
122–129
Publication date
31 March 2026
Language
EN
Document type
Book chapter