Homecare-Oriented Intelligent Long-Term Monitoring of Blood Pressure Using Electrocardiogram Signals

Long-term blood pressure (BP) monitoring is a widely used approach in a homecare intelligent system. However, BP is usually measured using cuff-based devices with tedious operation in practice, which may not be cost effective for continuous BP tracking. In this article, we propose a novel attention-based multitask network with a weighting scheme for BP estimation by analyzing and modeling single lead electrocardiogram (ECG) signals. Experimental results demonstrate that the proposed method could achieve mean error of systolic blood pressure, diastolic blood pressure, and mean arterial pressure estimation in levels of 0.18 ± 10.83, 1.24 ± 5.90, and 0.84 ± 6.47 mmHg, respectively. In comparison to other cutting-edge methods using ECG signals, the proposed method shows superior BP estimation performance. By integrating with a wearable/portable ECG monitoring device, the proposed model can be deployed to an embedded system or remote healthcare intelligent system to provide long-term BP monitoring service, which would help to reduce the incidence of malignant events happened in hypertensive population.