Human Activity Recognition and Assessment

The use of inertial measurement unit (IMU) sensors to monitor elderly mobility status is becoming increasingly important as the global population aged 60 years or over has doubled since the 1980s. Canada is one of the countries facing the highest increase of senior population, which requires healthcare institutions to explore novel solutions to tackle this aging-related challenge. Wearable IMU sensors represent a promising solution to accurately analyze the mobility and activity of older people, which is critical for primary care providers,  such as physicians and caregivers. To that end, a team of experts in deep learning and kinesiology have proposed a comprehensive framework that combines machine learning approaches with IMU and motion capture data to estimate three-dimensional knee joint kinematics. The objective is to predict gait kinematics of healthy adults and individuals with movement disorders with high accuracy and low cost. This approach has the potential to improve the traditional and expensive kinematic estimation procedures and become more accessible for clinical patients. The researchers aim to analyze data from IMU sensors worn by elderly people to provide primary care providers with an accurate assessment of their mobility status. To overcome the data shortage limitation and improve the performance of the machine learning model, the team proposed incorporating two learning paradigms: meta-learning and multi-task learning. Moreover, the researchers plan to simulate IMU data from motion capture and video data, which has proven to be abundant and a viable alternative to conventional IMU data collection. This approach has demonstrated up to 40% improvement in accuracy compared to the baseline model on various datasets. This project has the potential to lay a foundation for a comprehensive, flexible, and low-cost biomechanical data estimation framework for assessing gait in healthy adults and people with movement disorders. This approach would enable patients and clinicians to easily monitor movement and promote health, thereby reducing the burden of musculoskeletal disorders on the national healthcare system.