A cuffless blood pressure monitoring epidermal patch enabled by deep learning

In this work, a wrist-wearable epidermal patch comprising a piezoresistive ballistocardiogram (BCG) and an electrocardiogram (ECG) sensor for continuous and cuffless blood pressure (BP) monitoring has been developed and demonstrated. Specially designed electrode patterns were printed onto a flexible PET substrate for miniaturization and multimodal sensing. The BCG sensor showed a high sensitivity of -0.093 kPa(-1) at a low-pressure range (0-5 kPa). Simultaneous measurement allowed BP estimations through Convolutional Neural Network - Long Short-Term Memory (CNN-LSTM) based deep learning algorithm with an accuracy of MAE 7.798 for DBP and 3.540 for SBP which eliminated the need for traditional cuff-based systems.