Physical activity is one of the key contributors to health and quality of life. Running is popular and an efficient and affordable modality of physical activity. However, if done improperly, it may induce injuries leading to lower life quality and additional health and social costs. Therefore, it is important to provide tools for effective and injury-free physical activity. In the present study, a new generation of IMU sensors (smart) patches with considerably smaller dimensions and weight will be employed for rearfoot and pelvis stability measurement and their changes due to fatigue during running at different velocities and surface inclinations. This will be analysed through the pelvis and rearfoot motion patterns employing deep learning.

Results of the present study will show the eligibility for development of an on-line virtual running coach for safe running and for choosing the proper running shoes.
The overall objective of the project is to demonstrate a successful recognition of fatigue onset and
excessive pelvic and rearfoot mechanics at different running velocities and surface inclines using deep learning.

WP1: A software platform for capturing, storing, synchronization and processing of the captured data will be developed and will present an integral part of the project. The platform will be connected to the hardware data acquisition platform and will enable both automatic access to the data as well as access through a user interface. The majority of the platform will be located in the cloud, but its parts will also extend all the way to the acquisition hardware and computers on which the development will take place. Ultimately, the platform will be a key part of the demonstrators, where it will provide coaching based on an analysis in near real-time.

WP2: Conduct key measurements for project implementation needs. First, data will be collected by placing sensors on different locations to optimally capture motion patterns and their changes in deep learning. Then, measurements required for biomechanical analysis (WP3) and Deep Learning procedures and their verification (WP4) will be performed.

WP3: Provide information on the optimal placement of IMU sensors and signal conditioning to reliably track pelvic and heel movements, and set criteria for determining fatigue levels as input for deep learning.

WP4: Test state-of-the-art (SOTA) deep recurrent networks on the task of inferring expert biomechanical annotations from noisy data, obtained from wearable sensors. Adapt SOTA networks to improve performance on this particular problem.

WP5: Demonstrators will be developed to show the applicability of the newly developed machine learning procedures based on smart patch measurements (AiCoachU) in line with the main objective of the project.

AiCoachU – Artificial intelligence is coaching you (cobiss.net)