An ideally designed deep trust network model for heart disease prediction based on seagull optimization and Ruzzo...

Area of Science:

• Cardiology
• Artificial Intelligence
• Biomedical Engineering

Background:

• Heart disease risk factors include diet, stress, genetics, and sedentary lifestyles.
• Current automated diagnostic systems often focus on isolated aspects like feature selection or predictive accuracy.
• A comprehensive approach integrating multiple factors is needed to enhance cardiac prediction system efficiency.

Purpose of the Study:

• To introduce an ideally designed deep trust network (ID-DTN) for improved cardiac prediction system performance.
• To address network design challenges, overfitting, and lack of robustness in automated cardiac diagnostics.
• To optimize both network architecture and feature selection for enhanced system efficiency.

Main Methods:

• Utilized the Ruzzo-Tompa method for noncontributory feature elimination.
• Introduced the Seagull Optimization Algorithm (SOA) to optimize the deep trust network architecture.
• Evaluated the ID-DTN and restricted Boltzmann machine (RBM) performance using metrics like accuracy, F1 score, and Matthew's correlation coefficient.

Main Results:

• The proposed ideally designed deep trust network (ID-DTN) demonstrated superior performance compared to state-of-the-art methods.
• Achieved a significant prediction accuracy of 97.11% for cardiovascular disease.
• The integrated approach successfully optimized network architecture and feature selection.

Conclusions:

• The ID-DTN offers a more robust and efficient solution for cardiac prediction by considering multiple contributing factors.
• The study highlights the effectiveness of combining advanced deep learning techniques with optimization algorithms for medical diagnostics.
• The validated results provide a reliable foundation for developing advanced decision support systems for cardiovascular disease management.