Study on Model-based Estimation of Continuous Blood Pressure from Pulse Arrival Time using Two Doppler Radar Sensors

A unique method for extracting heartbeat and arterial pulse from noisy signals, based on mathematical models of heart and arterial movements, was proposed and evaluated for non-contact and continuous blood pressure (BP) estimation using pulse arrival time (PAT) for nine healthy participants, consisting of four females and five males. PAT was calculated from two Doppler radars positioned 0.5 meters away from the chest and inguinal sites. With this setup, conventional methods relying on low-noise signals failed to extract waveforms effectively due to weak and noisy radar signals. The proposed method generates templates of baseband signals of Doppler radar simulated from mathematical models, incorporating heart and arterial movements during systole and diastole. Using the proposed method, the Pearson correlation coefficient between heart rate and pulse rate was found to be a strong linear correlation of 0.99 (<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">p</i> < 0.01) compared to 0.96 (<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">p</i> < 0.01) obtained using the conventional method of peak detection. The standard deviations of the difference between estimated BP and measured BP were 9.93 mmHg for systolic BP and 10.51 mmHg for diastolic BP, where conventional peak detection methods proved unsuitable for BP estimation. During the Valsalva maneuver, estimated systolic BP exhibited changes consistent with typical BP variations, confirming the method’s potential for practical application.