CEBPM: A Cloud-Edge Collaborative Noncontact Blood Pressure Estimation Model

作者：Jia, Mengru; Qin, Yuting; Song, Cheng; Yue, Zijie; Ding, Shuai^*

来源：IEEE Transactions on Instrumentation and Measurement, 2022, 71: 5022712.

DOI：10.1109/TIM.2022.3205679

• Estimation
• Collaboration
• Videos
• Servers
• Cloud computing
• Monitoring
• Deep learning
• Attention mechanism
• cloud-edge collaboration framework
• deep learning
• noncontact blood pressure (BP) estimation

Blood pressure (BP) measurement is of great significance for the diagnosis of cardiovascular and chronic diseases. However, the existing BP measurement methods are uncomfortable to use and cannot provide continuous monitoring. To overcome these limitations, this article proposes a cloud-edge collaborative noncontact BP estimation model (CEBPM), which estimates BP from facial and palm videos within a cloud-edge collaboration framework. At the edge server, the feature extraction is conducted by the proposed special preprocessing procedure to improve the signal-to-noise ratio and reduce the network load for transmitting features. On the cloud server, the BP estimation network (BP-Net) is integrated with elaborately designed attention blocks (BPABs) and used for BP estimation in the online mode. Specifically, the attention block with temporal, spatial, and channel attention branches is introduced to obtain effective features. The experimental results demonstrate that the proposed model outperforms other competing methods of continuous BP estimation. The proposed collaborative noncontact BP estimation model can be leveraged in the regularization of BP monitoring in daily health management, especially for elderly and hypertension patients.