Applying Recurrent Neural Networks for Anomaly Detection in Electrocardiogram Sensor Data.

Area of Science:

• Cardiology
• Artificial Intelligence
• Biomedical Engineering

Background:

• Electrocardiograms (ECGs) are crucial for diagnosing heart conditions.
• Accurate ECG interpretation requires extensive training and expertise.
• Automating anomaly detection in ECGs can improve diagnostic efficiency.

Purpose of the Study:

• To explore the efficacy of recurrent neural networks (RNNs) for anomaly detection in ECG data.
• To optimize RNN architectures and training parameters for enhanced performance.
• To compare the performance of optimized RNNs against other contemporary optimization methods.

Main Methods:

• Recurrent neural networks (RNNs) were employed for anomaly detection in ECG signals.
• A modified particle swarm optimization (PSO) algorithm was used to tune network parameters and architectures.
• Performance evaluation involved comparing optimized RNNs with models from other optimizers.

Main Results:

• Optimized RNN models demonstrated significant potential for real-world application in ECG anomaly detection.
• The proposed optimization method yielded superior performance compared to contemporary optimizers.
• Feature importance analysis was conducted on the best-performing models.

Conclusions:

• Recurrent neural networks, when optimized, offer a promising approach for automated ECG anomaly detection.
• Particle swarm optimization effectively enhances RNN performance for cardiac diagnostics.
• This methodology has the potential to support clinical decision-making and improve patient care.