Cardiac arrhythmia detection using artificial neural network.

Area of Science:

• Biomedical Engineering
• Artificial Intelligence
• Wearable Technology

Background:

• Wearable medical devices face challenges in balancing performance with power consumption and cost.
• Advanced Artificial Neural Network (ANN) algorithms offer potential for sophisticated health monitoring in compact devices.
• Existing high-end wearable devices often have high power demands and costs.

Purpose of the Study:

• To develop a compact cardiac abnormality monitoring wearable device integrating ANN algorithms.
• To evaluate the feasibility and practicality of the Levenberg-Marquardt (LM) ANN algorithm for power-conscious wearable applications.
• To compare various neural network models for optimal performance in cardiac monitoring.

Main Methods:

• Development of a 'Cardiac Abnormality Monitoring' wearable medical device prototype.
• Integration and evaluation of the Levenberg-Marquardt (LM) ANN algorithm for embedded systems.
• Comparative analysis of six different neural network models, including Kernelized SVC with PCA.
• Assessment of performance, feasibility, and practicality through integration with a working prototype.

Main Results:

• The Levenberg-Marquardt (LM) ANN algorithm was evaluated for its suitability in power-constrained wearable devices.
• A comparative study identified 'Kernelized SVC with PCA' as a promising alternative model for testing accuracy.
• The performance and practicality of the ANN model were assessed via integration into a prototype device.

Conclusions:

• The developed wearable device demonstrates a viable approach to cardiac abnormality monitoring.
• The Kernelized SVC with PCA model shows potential for accurate and efficient cardiac monitoring in wearable technology.
• The study confirms the feasibility of using ANN algorithms in cost-effective, power-conscious wearable medical devices.