A Survey on A CNN Driven System for Early Age Heart Disease Prediction and Personalized Health Guidance

Abstract

Cardiovascular disease remains significant contributors to global mortality, with many cases progressing silently because of lack of noticeable symptoms during the initial stages. To address the early diagnosis, this study presents a novel approach employing a CNN based framework for the prediction of heart disease risk using electrocardiogram (ECG) image data. By analysing ECG patterns, the proposed model is capable of identifying potential indicators of cardiac abnormalities even in asymptomatic individuals. In addition to risk assessment, the system offers tailored health recommendations, guiding patients with personalized advice on lifestyle modifications, preventive measures, and further clinical evaluation. This functionality not only allows users to make informed health decisions. Furthermore, the model is integrated with an UI designed for accessibility and ease of use by both medical professionals and patients. This integration facilitates seamless interaction with the system, ensuring timely intervention and support in clinical settings as well as for individual health monitoring.

Introduction

• Cardiovascular diseases (CVDs) account for nearly one-third of global deaths.

• Rising cases among individuals under 40 are driven by sedentary lifestyles, poor diet, stress, and lack of early screening.

• Traditional diagnosis (ECGs, lab tests, clinical evaluation) is effective but inaccessible in rural/low-resource areas, leading to delayed diagnosis.

• The proposed solution aims to improve early detection and accessibility using AI technologies—especially Deep Learning (CNN) and Machine Learning (SVM).

2. Key Objectives

• Develop a hybrid AI system combining:

  • CNN for ECG image classification.

  • SVM for symptom-based diagnosis and medical advice.

• Provide early-stage detection, real-time interaction, and personalized recommendations—even without ECG input.

• Target young adults (20–40 years) and underserved populations to promote preventive cardiac care.

3. System Methodology

A. Data Collection and Preprocessing

• ECG datasets: MIT-BIH, PTB-XL, PhysioNet.

• Symptom datasets from UCI repositories or clinical records.

• ECG signals are denoised, normalized, segmented, and converted into 2D grayscale images for CNN input.

B. CNN-Based ECG Classification

• Custom CNN architecture identifies cardiac conditions like arrhythmia or ischemia from ECG images.

• Uses convolutional, pooling, and fully connected layers to extract and classify features.

C. SVM-Based Symptom Diagnosis

• Users input three symptoms.

• SVM model provides likely diagnosis and customized health advice, usable even without ECG access.

D. Integrated Decision Fusion

• If both ECG and symptoms are available, the system fuses CNN and SVM outputs to enhance accuracy.

• Weighted decision strategy ensures robust, actionable predictions.

E. Deployment and User Interface

• Available as a web/mobile app with a user-friendly interface.

• Backend (Flask/Spring Boot) handles inference and data processing.

• Designed for real-time, accessible use in both urban and rural settings.

4. Literature Comparison and Innovation

Innovation of Proposed System:

• First hybrid model integrating CNN-based ECG analysis with SVM-based symptom diagnosis.

• Focused on young adult heart disease, early detection, and accessibility.

5. Significance and Impact

• Bridges diagnostic gap in remote and under-resourced settings.

• Enables proactive and preventive cardiac care.

• Empowers individuals—especially youth—with early warning and personalized guidance.

• Promotes lifestyle changes and reduces long-term cardiovascular risks.

Conclusion

The system integrates CNN-based ECG analysis and SVM-driven symptom advice to enable early, accessible heart disease detection. It supports preventive care, especially for young and underserved populations.

References

[1] “ECG heartbeat classification: A deep transferable representation” – M. Kachuee, M. M. Kiani, H. Mohammadzade, M. Shabany – 2018 [2] “Personalized Monitoring of ECG Signals With Hybrid Feature Selection and Deep CNN Models” – S. Kiranyaz, T. Ince, M. Gabbouj – 2018 [3] “Pre selecting features for deep learning models: Improving convergence speed and reducing overfitting in ECG based heart disease detection” – J. Liang, J. Wang, Z. Zhang – 2020 [4] “Effective Heart Disease Prediction Using Hybrid Machine Learning Techniques” – S. Mohan, C. S. Thirumalai, G. Srivastava – 2019 [5] “Deep Learning Applications in ECG Analysis and Disease Detection: An Investigation Study of Recent Advances” – S. Udaya, K. K. Prakasha, S. Prabhu, V. C. Nayak – 2024 [6] “Automatic diagnosis of the 12 lead ECG using a deep neural network” – A. H. Ribeiro, M. H. Ribeiro, G. M. M. Paixão, et al. – 2019 [7] “Heart disease prediction using ECG based lightweight techniques” – A. Abbaszadeh, et al. – 2024

Copyright

Copyright © 2025 Sheela Rani C M, Naboday Saha, Jnanaskanda K, Lasya R Gowda, Nayan S. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.