Host university: Northumbria University
PhD research project: Understanding fall risk in real-world settings for people with Parkinson’s disease (U-Fall PD): Deep Learning Approach to Video Based Environmental Classification
Project Summary: Current clinical approaches to determining falls risk in PD are subjective and lack objective (bio) markers. However, advancements in technology have the potential to provide a more detailed understanding of an individual’s falls risk. One promising area of investigation is the use of wearable devices equipped with inertial sensors, which can capture gait (walking) data during daily activities. Those devices can be cost-effective while providing clinically valid gait (bio) markers indicative of fall risk. By analysing gait biomarkers, it is possible to tailor interventions to reduce falls in PD.
Nevertheless, a major limitation in the use of inertial-based wearables for gait assessment is the lack of contextual information. For example, although inertial wearables can be worn anywhere, clinicians do not know if the person was walking indoors or outdoors which is critically important for gait analysis in fall risk assessment. However, the use of wearable cameras has shown promise in providing contextual information but remains largely underexplored. Accordingly, the combination of inertial sensors and video analysis in a clinically driven manner is yet to be achieved.
My research project aims to investigate artificial intelligence (AI) based computer vision with inertial wearables for a complete gait assessment in the home. My work (i) ethically analyses videos from free-living environments to (ii) provide valuable insights into gait assessment for improved fall risk assessment in PD. A novel aspect of my work is the use of wearable eye-tracking video glasses, to provide contextual video data as well as understanding eye movements to understand how individuals with PD navigate their environments, such as walking in crowded spaces or navigating stairs. The findings will inform the development of tailored strategies to reduce falls, enhancing safety and independence for people with PD.
What is this project about?
Context aware free-living environment gait analysis within people with Parkinson’s Disease (PwPD)
What is the aim of this project?
To develop computer vision-based AI and machine learning algorithms to automatically provide context to inertial sensor data stemming from free living environments.
Who will benefit from this project?
People with Parkinson’s disease who experience falling.
Expected PhD completion date