Blood pressure class estimation using CNN-GRU model

Abstract

Blood pressure (BP) estimation can add on great value in emergency medicine, especially in case of mass casualty situations. The presented research aims to create a model for BP class estimation using electrocardiogram (ECG) and photoplethysmogram (PPG) waveforms. We focus on developing a BP classification model as a convolutional neural network (CNN) - gated recurrent unit (GRU) hybrid model, containing both CNN and GRU layers. The used dataset is the publicly available UCI Machine Learning Repository dataset. We have achieved f1 score of 0.83, 0.73 and 0.74 respectively according to the BP classes and 78% overall accuracy.