Improved Neural Network Arrhythmia Classification Through Integrated Data Augmentation

This work investigates an evolution of verified recent advances to machine learning applied to electrocardiogram (ECG) data. The successful inference of heartbeat arrhythmia has long been a goal yet achieved, the techniques presented advance the worthy endeavor. The mutation of the training data through amplitude and time inversion creates artificial information leading to a more robust and accurate model in comparison to the current state of the art. Over a 5% reduction in accuracy error is reached with the proposed techniques in comparison to that of the base model.