Atrial fibrillation classification using deep learning algorithm in Internet of Things-based smart healthcare system

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Signal Processing

Background:

• Detecting electrocardiogram (ECG) patterns in Internet of Things (IoT)-based healthcare systems presents significant challenges.
• Accurate classification of ECG signals is crucial for timely diagnosis and patient notification.
• Advanced computing methods are essential for improving ECG signal analysis in e-health applications.

Purpose of the Study:

• To propose an intelligent ECG signal classification method using a deep learning algorithm for IoT-based smart healthcare.
• To develop and test a partitioned deep convolutional neural network (pCNN) for classifying ECG signals.
• To evaluate the performance of the proposed pCNN against conventional classifiers for atrial fibrillation detection.

Main Methods:

• Acquisition of continuous ECG features within an IoT-based monitoring system.
• Development of a partitioned deep convolutional neural network (pCNN) for ECG signal classification.
• Comparative analysis of the pCNN with other classification algorithms using time-series data from atrial fibrillation samples.

Main Results:

• The proposed pCNN achieved an accuracy of 96.3%, sensitivity of 93.5%, and precision of 97.5% in classifying atrial fibrillation.
• Learned features from the pCNN demonstrated reliable performance when tested against conventional classifiers.
• The pCNN effectively utilized learned features in continuous time-series data by forming a higher-order space on the server.

Conclusions:

• The partitioned deep convolutional neural network offers a reliable and accurate method for ECG signal classification in IoT-based smart healthcare systems.
• The proposed deep learning approach provides timely assistance and enhances diagnostic capabilities for conditions like atrial fibrillation.
• This research highlights the potential of advanced deep learning algorithms in advancing e-health monitoring and analysis.