An Automatic Cardiac Arrhythmia Classification System With Wearable Electrocardiogram

摘要: This paper presents an automatic wearable electrocardiogram (ECG) classification and monitoring system with stacked denoising autoencoder (SDAE). We use a device wireless sensors to obtain the ECG data, send these data computer Bluetooth 4.2. Then, are classified by cardiac arrhythmia system. First, feature representation is learned SDAE sparsity constraint. softmax regression used classify beats. In fine-tuning phase, active learning added improve performance. we method that relies on deep neural networks posterior probabilities associate confidence measures select most informative samples. Breaking-ties modified breaking-ties methods validate proposed well-known MIT-BIH database obtained from device. follow recommendations of Association for Advancement Medical Instrumentation class labeling results presentation. The show performance our approach outperforms state-of-the-art methods.