Development of machine learning technique for wearable vital sign monitoring device

• Main Author: Zheng, Yandan
• Other Authors: Muhammad Faeyz Karim
• Format: Theses and Dissertations
• Language: English
• Published: 2019
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Antennas, wave guides, microwaves, radar, radio
• Online Access: http://hdl.handle.net/10356/78810
• Institution: Nanyang Technological University

The Wearable monitoring device, especially the non-contact wearable monitoring device is prevalent recently. It can monitor the vital signs (heart rate and respiration rate) continuously which can be useful in several situations — for example, athletes body condition monitoring and patient emergency alert. Most of the non-contact device embed with radar that can perform monitoring task in the distance. Millimeter wave based Frequency Modulated Continuous Wave (FMCW) radar can detect human motion in millimeter sense. However, numerous outer factors can affect sensor performance. In the digital world, machine learning can adapt to different data into different input factors and hence perform future correction. In real life, usually, there is only real-time discrete ground truth for the machine learning training process. To tackle the problem in real life scenarios, this work proposed Semi-supervised regression-based vital sign learning technique with co-training and ensemble learning style. The final trained model achieves excellent results.