USING THE TIME-DOMAIN CHARACTERIZATION FOR ESTIMATION CONTINUOUS BLOOD PRESSURE VIA NEURAL NETWORK METHOD

Abstract

The new method with back-propagation neural network is expected to be capable of continuous measurement of blood pressures with noninvasive, cuffless strain blood pressure sensor. The eight time-domain characterizations estimate systolic blood pressure and diastolic blood pressure via BPNN leading to a satisfactory accuracy of the BP sensor. The BP sensor is used on human wrist to collect the continuously pulse signal for measuring blood pressures. To assist the sensor, a readout circuit is devised with a Wheatstone bridge, amplifier, filter, and a digital signal processor. The results of SBP and DBP are 4.27 +/- 4.98 mmHg and 3.86 +/- 5.35 mmHg, respectively. The errors of blood pressure pass the criteria for Association for the Advancement of Medical Instrumentation (AAMI) method 2 and the British Hypertension Society (BHS) Grade B.