Data Analytics of BP-PPG Dataset: Noninvasive Blood Pressure Assessment by Using Photoplethysmography Fiducial Points

One of the most important features of a monitoring medical device is the ability to track changes in a non-invasive manner. In the last decade, the focus moved on photoplethysmography technique (PPG) which can be the base for a such device for monitoring noninvasively blood pressure (BP). Although the PPG and BP recorded signals outsourced from the cardiovascular system the two types of recording present the same signal morphology even if is by optical or mechanical measurement. Before extracting physiological data with the PPG method a feasibility study needs to be done in order to verify if the mechanical information contained in the BP signal is conserved at an acceptable degree within the optical one. For this study, the relevant data is given by BP-PPG dataset pairs of MIMIC III database which are collected from over 12.000 patients. The results obtained show that fiducial points of BP values are correlated with PPG with a factor of 0.47 in many cases, but also shows that interdependency between SYS, DIA, and pulse pressure rank the correlation results. This is the effect of various biological effects as waves traveling through elastic medium and vasomotor phenome which occurs in the cardiovascular network. Therefore, the PPG signal should be studied from more advanced points of view in order to find the conserved mechanical information.