Arrhythmia detection based on patient-specific normal ECGs using deep learning

Most traditional studies regarding arrhythmia detection using electrocardiogram (ECG) have proposed general methods applicable to various patients. Because patients have their own unique ECG patterns, abnormalities undetected by general methods can be detected if a new arrhythmia detection method tailored to each patient is developed. Furthermore, the new method can effectively support doctors in their diagnosis if it can provide the basis for determining abnormalities. Herein, we propose an individualized ECG abnormality judgment method using an autoencoder and a convolutional neural network. This method allows the autoencoder to learn only normal waveforms that can be easily collected and obtains the characteristics of the individual's unique normal waveforms. Our method compares the features acquired from the ECG pattern to be analyzed with those of the normal waveform and determines whether they are normal or abnormal. In addition, we aim to construct a system that can demonstrate the basis for the judgment of whether a feature is normal or abnormal by showing the acquired features.