Related Experiment Video
Updated: Jun 30, 2026
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
Using convolutional neural networks with late fusion to predict heart disease
Deema Mohammed AlSekait1, Mohammed Zakariah2, Syed Umar Amin3
1Department of Information Technology, College of Computer and Information Sciences, Princess Nourah Bint Abdulrahman University, P.O. Box 84428, 11671, Riyadh, Saudi Arabia.
Scientific Reports
|November 21, 2025
Summary
This study introduces a novel deep learning method using convolutional neural networks (CNNs) and deep neural networks (DNNs) for accurate heart disease prediction. The hybrid model achieved near-perfect scores, enhancing patient health and medical data management.
Area of Science:
- Cardiology
- Artificial Intelligence
- Machine Learning
Background:
- Cardiovascular diseases (CVDs) are a leading global cause of death, necessitating improved diagnostic tools.
- Machine learning and data mining offer advanced methods for faster and more accurate patient diagnosis.
- Traditional diagnostic methods can be time-consuming and may lack the precision of AI-driven approaches.
Purpose of the Study:
- To present a novel late fusion deep learning method for predicting heart disease.
- To develop a hybrid architecture combining CNNs and DNNs for enhanced diagnostic accuracy.
- To improve patient health outcomes and streamline medical data management through precise prediction algorithms.
Main Methods:
- Utilized a dataset of 303 instances and 13 features from the UCI Machine Learning Repository.
- Implemented a late fusion technique combining Convolutional Neural Networks (CNNs) for specialized data modalities and Deep Neural Networks (DNNs) for tabular data analysis.
- Developed a hybrid architecture merging numerical features with graphical representations to capture spatial and sequential properties.
Main Results:
- The novel late fusion model demonstrated high accuracy in identifying heart illnesses.
- Achieved zero percent error in accuracy, precision, and recall on validation and testing sets.
- Attained an F1 score of 99.99%, indicating exceptional predictive performance.
Conclusions:
- The developed hybrid deep learning model offers a precise and extensible solution for heart disease prediction.
- This approach significantly enhances the accuracy of medical diagnostics, providing a strong foundation for future research.
- The findings contribute to improved patient care and more efficient medical data management.
