Hypertensive patients account for about 16% to 37% of the global population, and about 9.4 million people die each year from hypertension and its complications. Blood pressure is an important indicator for diagnosing hypertension. Currently, blood pressure measurement methods are mainly based on mercury sphygmomanometers in hospitals or electronic sphygmomanometers at home. However, people's blood pressure changes with time, and using only the blood pressure value at the current moment to judge hypertension may cause misdiagnosis. Continuous blood pressure measurement can monitor sudden increases in blood pressure, and can also provide physicians with long-term continuous blood pressure changes as a diagnostic reference. In this article, we design an artificial intelligence-enhanced blood pressure monitoring wristband. The wristband's sensors are based on piezoelectric nanogenerators, with a high signal-to-noise ratio of 29.7 dB. Through the transformer deep learning model, the wristband can predict blood pressure readings, and the loss value is lower than 4 mmHg. By wearing this blood pressure monitoring wristband, we realized three days of continuous blood pressure monitoring of the subjects. The blood pressure monitoring wristband is lightweight, has profound significance for the prevention and treatment of hypertension, and has wide application prospects in medical, military, aerospace and other fields.
Keywords: artificial intelligence; biosensors; deep learning; piezoelectric nanogenerator; self-powered.