Neural ECG Beat Analysis for Noise-Robust Cardiac Detection

The patent introduces an intermediary neural network model that acts as a mediator between the noisy ECG signal and the cleaned signal. The model includes a generator that creates cleaned signals and a discriminator that verifies their authenticity, ensuring that critical cardiac information is preserved while removing noise. This intermediary system avoids the direct signal processing that causes information loss in traditional methods.

The patent replaces traditional mechanical de-noising techniques (such as filtering algorithms) with a neural network-based system. The neural network learns the complex patterns of cardiac signals and noise through training, substituting deterministic mechanical filtering with adaptive learned filtering that preserves critical information while removing noise.

The patent applies preliminary action by training the neural network model offline before deployment. The model learns from large datasets of ECG signals during the training phase, capturing complex patterns and relationships. Once trained, the model can make accurate real-time predictions with reduced computational burden during actual ECG analysis, as the heavy lifting of pattern recognition has already been performed during training.

The patent segments the ECG signal into individual cardiac cycles or beats for analysis. This segmentation allows the neural network to process smaller, more manageable units rather than entire long-duration ECG recordings, reducing computational complexity while maintaining measurement precision. Each segmented beat can be independently analyzed and classified.