Evaluation of One-Point Calibration for Cuffless BP Wearable Sensor Devices: Stiffness Index

Blood pressure (BP) is one of the most important vital signs for the early diagnosis of cardiovascular diseases (CVDs). This article presents a noninvasive continuous BP measurement system using one-point calibration. The conventional continuous BP measurement in wearable sensor devices uses pulse arrival time (PAT) or pulse transient time (PTT), which requires multiple measurements for calibration. In addition, users need to increase their BP through exercising for calibration. In this study, we developed a novel algorithm that only requires one steady BP measurement for calibration. This calibration method is based on the population calibration parameters and their artery stiffness information from photoplethysmography (PPG) signal. The single calibration method uses a two-stage linear regression approach based on people’s arterial stiffness index (SI) for wearable devices. Methods: PPG and electrocardiogram (ECG) sensors are attached to the hands to measure analog signals. Simultaneously, a cuff device is attached to the right arm as a reference. After raw PPG and ECG signals are processed with filters, landmarks can be successfully extracted using our proposed algorithm. We have tested the data with different body postures (sitting, standing, and supine) and varying levels of exercise intensity. We collected and analyzed data from 41 individuals, including 30 males and 11 females for this study. Finally, individual and population linear regression methods are applied to the data to reduce the calibration process. The results of this study give an average mean absolute error (MAE) of 3.832 and an average
${R}^{{2}}$
of 0.8034 for systolic BP (SBP). About 87.33% of data are within the accuracy of ±10 mmHg.