Research On Key Technologies For Combined Monitoring Of ECG And Blood Pressure

With the increasing pressure of modern life, the incidence of cardiovasculardisease shows an increasing trend. Coronary heart disease may lead to myocardialinfarction, heart failure, sudden death and other serious consequences. Therefore themonitoring of coronary heart disease is very important to cardiovascular patients.Currently, the most of monitoring systems on the market for coronary heartdisease denpends on ECG signal only. Due to the fact that the ECG signals have thecharacteristics of complexity and variety, the stability and accuracy of these systemscan not satisfy the clinical requirement. To reduce the occurs of wrong prediction, thispaper proposed a novel monitoring system.The main principle of this system is that, many research have shown thatcoronary heart disease and hypertension are closely related. Moreover, hypertensionpatients are more likely dying from coronary disease than normal patients. In order toimprove the diagnostic accuracy, we combined the measurements of blood pressureand ECG signal together. The main works of this paper are as follows:Fisrtly, we designed a combined monitoring system of ECG signal and bloodpressure, which monitors the patient’s ECG and processes in real-time. A severeabnormality will trigger the blood pressure measurement.Secondly, we designed a wrist blood pressure measurement module, in which theblood pressure was computed based on gaussian fitting and oscillometric methods.And then the accuracy of the module was validated through comparison to OMRONblood pressure meter. The results showed that the relative error was less than5%,which indicated that the stability and accuracy of the designed module were in anaccepted range. Thirdly, a novel ECG classification algorithm was proposd, which can beimplemented on mobile platforms. To improve the accuracy of QRS detection, awavelet transform based method was illustrated. Additionally, different fromtraditional modeling method like BP neural network, support vector machine etc., theextreme learning machine (ELM) was applied to reduce the complexity and modelingtime. The experiment of MIT-BIH database showed that the proposed classificationalgorithm can precisely identify the abnormal ECG signals from normal ECG signals.Finally, to solve the nonlinearity problems this paper introduced a modified ELMalgorithm, in which the traditional nonlinear transfer functions of hidden neuronswere replaced by the combination of linear and nonlinear functions. Additionally, toavoid the collinearity problem, the ridge regression method was introduced into theELM algorithm. The results indicated that the modified ELM algorithm was moresuitable for classification of ECG abnormalities.