Pulse Wave Propagation Model Based On Waveform Separation And Its Cloud Applications

Pulse wave contour analysis technology has been used for cardiovascular function evaluation for a long time, given its convenience and high reliability. With the deepening research of hypertension and coronary heart disease, more effective methods are needed.In clinical applications, pressures pulse wave detection for artery and photoplethysmography(PPG) detection for microvascular. Pressure pulse wave has definite physiological significance, although higher operating skills and more expensive devices are required. While PPG signal has strong stability and can be detected conveniently. However, as the origins of the different components of the PPG signal are still not fully understood, and PPG signal doesn’t behave like pressure pulse, models traditional used to assess cardiovascular statement shows poor effectiveness.Consequently, the relationship between finger PPG and radial pressure pulse were researched. In this study, Gaussian modeling method was used to obtain radial pressure waveform from finger PPG. This also led a new way to observant waveform variation in pulse wave propagation.In this thesis, finger PPG and radial pressure pulse signal were recorded from 200 subjects non-invasively. For each subject, finger PPG and radial pressure pulse signal were acquired simultaneously. Then pulses were normalized beat-by-beat for both width and amplitude and then averaged to obtain a normalized reference pulse waveform for finger PPG and radial pulse respectively. The two reference waveforms from each subject were modeled using three Gaussian functions, with nine parameters, including the peak amplitude, peak time and half-width from each Gaussian function. Nine Gaussian parameters of finger PPG add to basic clinical information as input features to multivariate regression models to estimate each of Gaussian parameters of radial pulse one by one. Then substituted estimated radial parameters into expression, radial pulse waveform can be reconstructed. 10-fold cross validation was employed to assess the model performance. These promising results indicate the feasibility of using Gaussian modeling method to obtain radial pressure waveform from finger PPG.In this way, high reliability of radial pulse wave are reconstructed by finger PPG, and physiological significance associated with Gaussian functions were also explored.Noting high performance requirements for computing, the model was attempted to realize on cloud platform, to support portable devices. A Django Web services framework base on Python was used. And some advices of healthcare server on Internet were given at last.