Posture-related back pain degrades mobility and quality of life for many who experience it. Building a healthy posture can reduce strain on the spine and help avoid these back problems. However, recognizing and correcting one's posture can be difficult to accomplish on one's own. In this work, a wearable technology solution is presented. Using a T-shirt with embedded flexural strain sensors, this work enables users to recognize when they have bad posture and correct it in real time. To extract useful information from the flexural strain sensors, a machine learning algorithm was trained to utilize the flexural strain sensor output to predict whether the user had good posture, leveraging existing vision-based algorithms to extract human pose. The model demonstrated high accuracy, showing the viability of this technology. Overall, this work presents a reliable, non-intrusive solution for monitoring and correcting bad posture, which can provide valuable feedback and reduce overall back-related problems.
Zahoori et al. (Sat,) studied this question.