Research On Key Issues In Wearable ECG Monitoring

The widespread cardiovascular disease(CVD)has become the "number one killer" endangering human life and health.The wearable electrocardiogram(ECG)monitoring system is an effective technical means to solve the early screening and real-time monitoring of CVDs,which can significantly reduce the mortality/disability rate of CVDs and reduce social and economic losses.However,the complex motion conditions in the wearable environment pose new challenges to the performance of dry electrodes,then it is necessary to study the influence of the dry electrode on the signal link.Meanwhile,severe CVDs require multi-lead wearable ECG monitoring systems for accurate diagnosis,so it is urgent to strengthen the research on wearable ECG lead selection and optimization related technology;In addition,for the real-time-based and cloud-based abnormal heartbeat identification requirements,the development of the corresponding abnormal heartbeat classification algorithms have become an urgent problem to be solved.In this study,these challenges are investigated and discussed from the perspective of signal acquisition and processing,including dry electrode test and system prototype design,lead optimization and abnormal heartbeat recognition.The main research contents and progress of this study are as follows:(1)In view of the influence of various factors on the performance of the dry electrode in the signal chain under wearable environment,the influence of dry electrode on skin-electrode impedance is investigated,and a continuous,non-invasive and comfortable wearable ECG monitoring system is proposed based on the dry electrodes and flexible printed circuit board(FPCB)-based ECG processing module;(2)In order to design and optimize the lead configuration for the multi-lead wearable ECG monitoring system,a myocardial infarction(MI)-region-refined torso-heart model is proposed to explore the effect of inferior MI on simulated ECGs,the results show that aVF lead has significant advantages in reflecting the size and location of inferior MI,and this method provides a lead optimization guidance for the prospective medical application of multi-lead wearable ECG monitoring;(3)Aiming at the problem of limited computing resources of the embedded analysis platform of the wearable ECG monitoring system,a rule-based real-time premature beat(PB)recognition algorithm and a rule-based long-term PB recognition algorithm are designed to realize the classification of normal(N),premature atrial beat(PAC)and premature ventricular beat(PVC)in single-lead wearable ECG data,the total recognition accuracy of the real-time PB detection algorithm is 97.51%,and long-term PB recognition algorithm has been verified in clinical data;(4)For cloud computing process of massive ECG data,a two-dimensional PB recognition deep learning model based on AlexNet-like network and a one-dimensional PB recognition deep learning model based on time-series framing network are proposed,these models are trained on noisy clinical data and tested on wearable data,and the recognition accuracies of these two models are 89.33% and 89.73%,respectively.This study has further enriched the research on signal perception and abnormal recognition in the field of wearable ECG monitoring.It provides new ideas and technical support for wearable ECG signal acquisition and monitoring,early disease screening and intelligent diagnosis and evaluation.