Wearable sensing system to perform realtime 3D posture estimation for lower back healthcare

Wearable sensing devices have a huge potential in developing innovative solutions for healthcare problems. Lower back or lumbar-pelvic movement monitoring is proven to be an effective method in tackling back pain problems. Significant applications of wearables in postural control therapies are crucial in the prevention of low back pain. These methods reduce costly hospitalization and allow the self-rehabilitation process. This research paper introduces a novel wearable sensing system to perform real-time 3D posture estimation for lower back healthcare. The real-time 3D system is divided into three components: a wearable device at the lumbar region; a wearable watch device; and a computer vision unit. These devices use BLE (Bluetooth Low Energy) technology for wireless communication. The wearable units utilise inertial measurement unit (IMU) sensors to perform position and orientation measurements of the body. The motion sensors are compact, cost-effective, and low power, which makes them more accessible and useful. The sensing system is developed and calibrated using practical data and cross-verified using real-world datasets. The computer vision system effectively estimates the human posture using enhanced machine learning algorithms with >97% accuracy. The research focuses on estimating the correct postures using dual-IMU and computer vision data. The paper conclusively proposes the development of a DNN based approach in analysing and improving body posture accuracy.