ECG-Based Blood Pressure Estimation Using a Two-Stage Inception-Regression Model

Abstract

For patients with Cardiovascular Diseases (CVDs), their blood pressure (BP) and electrocardiograms (ECGs) have to be monitored daily in order to avoid any complications and increase their quality of life. This paper presents a novel BP estimation approach using ECG signals, based on a two-stage inception model. The first stage employs an inception-based convolutional neural network (CNN) to automatically extract ECG features correlated with BP. These features are then processed by a regression algorithm to predict systolic (SBP) and diastolic (DBP) blood pressure. The model was trained using the MIMIC-II dataset and demonstrated superior performance in estimating both DBP and SBP, outperforming state-of-the-art methods.