Transfer Learning Enhanced Blood Pressure Monitoring Based on Flexible Optical Pulse Sensing Patch

Abstract

Blood pressure (BP), a crucial health biomarker, is essential for detecting early indications of cardiovascular disease in routine monitoring and clinical surveillance of inpatients. However, conventional cuff-based BP measurements are limited in providing continuous comfort monitoring. Here, we present an optical pulse sensing patch for BP monitoring, which integrates three units of Gallium Nitride (GaN) optopairs with micronanostructured polydimethylsiloxane films to capture pulse waves. Multipoint pulse signals are transformed into BP and other cardiovascular indicators through machine learning. The transfer learning method is developed to calibrate the machine learning model with few training sets, simplifying the practical implementation. The developed sensing patch holds great potential for long-term, precise BP monitoring, enhancing clinical diagnosis, and management of cardiovascular diseases.