A blood pressure measurement system based on internet of things and deep learning is proposed for continuous data acquisition and blood pressure prediction.The system adopts a hybrid neural network structure for processing the collected data and accurately predicting blood pressure,and the model consists of ResNet18,GRU and 3 fully connected layers.The data of 82 individuals are collected for training and testing.The mean absolute errors and standard deviations are 2.16 mmHg and 3.09 mmHg for diastolic blood pressure,3.15 mmHg and 5.14 mmHg for systolic blood pressure,according with AAMI standard and BHS standard.
Result Analysis
Print
Save
E-mail