A Congestive Heart Failure Detection System via Multi-Input Deep Learning Networks

摘要

In this study, a detection system of congestive heart failure (CHF) based on multiple input neural network was proposed. In previous research, the majority of studies focused on 24-hour electrocardiogram (ECG) data analysis for classification. To provide a convenient and rapid screen process, we proposed to offer a short- term analysis with the data size of 7-minute segment of ECG signal. The proposed detection system consisted of four steps: data pre- processing, model- establishment, multi-input configuration, and deep learning model classification. We proposed RR intervals instead of raw ECG data for the model input to reduce computation complexity. Also, by feeding in RR interval signal in both time and frequency domain, we can leverage the model performance by the known study results from HRV analysis to obtain the significant features more easily. The recognition accuracy between CHF and control groups of proposed detection system is up to 93.76% for training set, and 86.74% for testing set.