CNN and SVM-Based Models for the Detection of Heart Failure Using Electrocardiogram Signals

Area of Science:

• Biomedical Engineering
• Cardiology
• Artificial Intelligence in Medicine

Background:

• Heart failure (HF) significantly impacts health, necessitating early and accurate detection for effective management.
• Electrocardiograms (ECGs) are crucial for identifying conditions that may lead to HF, but interpretation can be challenging and time-consuming.
• Automated detection of HF from ECGs can overcome interpretive limitations and improve diagnostic efficiency.

Purpose of the Study:

• To develop and evaluate two novel AI models for the automatic detection of heart failure (HF) from ECG signals.
• To compare the performance of a Convolutional Neural Network (CNN) model with a CNN-SVM hybrid model for HF detection.
• To assess the efficiency and accuracy of these models using standard cardiac databases.

Main Methods:

• Two distinct models were proposed: a Convolutional Neural Network (CNN) and a CNN integrated with a Support Vector Machine (SVM) layer.
• The models were trained and tested using the MIT-BIH and BIDMC ECG databases, utilizing 2-second ECG fragments.
• Model architecture was optimized for reduced training time and memory consumption compared to existing methods.

Main Results:

• Both proposed models demonstrated high effectiveness in automatic HF detection.
• Achieved accuracy, sensitivity, and specificity exceeding 99% through blindfold cross-validation.
• The models are computationally efficient, requiring less training time and memory.

Conclusions:

• The developed AI models offer a highly accurate and efficient method for automatic heart failure detection from ECGs.
• These models can serve as valuable tools for clinicians, providing reliable diagnostic support.
• The models' efficiency makes them suitable for integration into portable devices for real-time patient monitoring.