ECG Classification Algorithm Based on STDP and R-STDP Neural Networks for Real-Time Monitoring on Ultra Low-Power...

Area of Science:

• Biomedical Engineering
• Artificial Intelligence
• Wearable Technology

Background:

• Real-time cardiac monitoring is crucial for managing cardiovascular health.
• Existing systems often face limitations in power consumption for wearable devices.
• Spiking neural networks (SNNs) offer a promising, low-power alternative for signal processing.

Purpose of the Study:

• To develop a novel, ultra low-power electrocardiogram (ECG) classification algorithm.
• To integrate this algorithm into real-time cardiac monitoring systems for wearable devices.
• To evaluate the algorithm's accuracy and energy efficiency compared to existing methods.

Main Methods:

• Implementation of a novel algorithm based on spiking neural networks (SNNs).
• Utilized spike-timing dependent plasticity (STDP) and reward-modulated STDP (R-STDP) for model training.
• Trained model weights based on spike signal timings and reward/punishment signals.

Main Results:

• The proposed SNN-based algorithm is suitable for real-time operation.
• Achieved accuracy comparable to previous neural network-based ECG classification methods.
• Demonstrated significantly lower energy consumption: 1.78 μJ per beat, 2-9 orders of magnitude less than prior methods.

Conclusions:

• The developed SNN algorithm offers a highly energy-efficient solution for real-time ECG classification.
• This technology is well-suited for integration into ultra low-power wearable cardiac monitoring systems.
• The findings pave the way for more sustainable and effective remote cardiac health management.