Cuff-Less Non-Invasive Blood Pressure Measurement Using Various Machine Learning Regression Techniques and Analysis

Top

1. Introduction

Blood Pressure (BP) is the force excreted by the blood on the blood vessels and arteries during circulation. The pressure reduces as the blood flows away from the heart. BP is one of the most vital parameters studied and analyzed in medical and health care systems. The blood pressure varies with every heartbeat and the pressure is highest when the heart contracts and pumps the blood to arteries, known as Systolic Blood Pressure (SBP). During the relaxation interval of the heart, the pressure is least and it is known as Diastolic Blood Pressure (DBP). A wearable physiological monitoring system developed by Pandian P S et al (2008) named ‘Smart Vest’, measures and transmits biomedical signals like ECG, PPG, GSR, blood pressure and body temperature to a remote monitoring system. The heart rate is derived from the ECG signals. The systolic and diastolic blood pressures are derived non-invasively using Pulse Transit Time technique. At the remote base station non-invasive BP computation is carried out using the calibration equation and the trend analysis is done. Heiko Gesche et.al (2011) have proposed that there is a correlation between pulse wave velocity (PWV) and SBP. The aim of the study was to develop a nonlinear algorithm and a one point calibration for the measurement of SBP. It was found that SBP calculated from PTT with the reference BP measured from cuff-based device had a correlation in the range of an empherical formula was developed to calculate the SBP. Yan et.al (2007) proposed a novel calibration method for non-invasive BP measurement using pulse transit time technique. In their work they explained that PTT and BP are linearly related by the equation where a & b are constants. Assuming that the pressure under arm cuff linearly decreases across the artery then:

where L₁ & L₂ represents cuff width and the artery length from the cuff to the finger, P₀ is the mean BP at the heart and is the internal pressure drop under cuff. With these calibration equations they calculated the BP non-invasively. Wan Suhaimizan Zaki et.al (2016) have developed a system for continuously monitoring the blood pressure without using the cuff. In this method, two types of PTT were measured. In the first method PTT₁ was measured at the fingertip and the PTT₂ was measured at the brachial artery. The experiment showed that PTT₁ has good correlation with BP whereas PTT₂ has weak correlation.