Using wearable sensors and deep learning for impact activity assessment during sports

Injuries are common in sports activities and can have a severe impact on an athlete's physical health and future development. Traditional methods for healthcare monitoring, such as manual video analysis, have limitations in real-time feedback and accurate analysis. Deep learning methods such as time-series analysis models offer a promising approach to overcome these limitations by training on the data collected by wearable sensors, that has the potential to provide real-time insights into the athlete’s health state monitoring. The research aims to explore the feasibility of utilizing unobstructive wearable sensors and machine learning/deep learning methods to assess impact activities by time-series analysis during sports, such as football or volleyball, thereby it is able to mitigate injury risks for athletes and help adjust training arrangements from the perspective of coaches. The expected project output includes a robust assessment system incorporating proposed methods, a public repository of code output and dataset for future research, and several publications to summarize the results.