Blood pressure monitoring via double sandwich-structured triboelectric sensors and deep learning models

Real-time blood pressure monitoring is essential for the timely diagnosis and treatment of cardiovascular disease. Many traditional prediction methods estimate blood pressure by measuring multiple sets of physiological signals with energy-consuming sensors. Herein, a continuous, cuffless and self-powered blood pressure monitoring system was developed based on a new double sandwich-structured triboelectric sensor and a novel blood pressure method estimation. A pyramid-patterned sensor based on the double sandwich structure realizes a sensitivity of 0.89 V/kPa in a linear range of 0–35 kPa, which is more than twice of the conventional single electrode structure. The sensor processes a low pressure detection limit of 1 g andfast response time of 32 ms. Hence, it can easily capture the pulse signal at the radial artery. Furthermore, a novel method for estimating blood pressure using pulse waves accompanied by the user’s background information was proposed. This method measures only one set of pulse signals and is portable. A deep learning model with multi-network structures was developed to improve the estimation accuracy. The mean absolute error and standard deviation of error for systolic and diastolic blood pressure (SBP and DBP) estimations were 3.79 ± 5.27 and 3.86 ± 5.18 mmHg, respectively. This work reveals a new sensing structure of triboelectric sensors and offers a novel method for blood pressure estimation.