Using artificial intelligence to personalise exercises for people with low back pain
Mae'r cynnwys hwn ar gael yn Saesneg yn unig.
Low back pain (LBP) is a global leading cause of disability affecting 30-50% of people (28 million) in the UK at any one time.
The majority of LBP has no specific pathology that can be cured or medicated. Instead, clinical guidelines recommend active physical interventions, in other words exercise tailored to individual needs.
LBP is complex and multifactorial. Individually tailoring exercises is a major clinical challenge. Artificial intelligence (AI) holds a major promise in individualising management of complex health conditions including LBP. This project will use AI to build a platform that delivers rehabilitation exercises bespoke to an individual’s underlying physical condition.
Current systems (such as Kaia Health, Hocoma) guide a person to achieve a “gold standard” exercise performance, not considering movement problems or physical de-conditioning. AI currently cannot offer the bespoke exercise programmes that are required tand are therefore unsuitable (potentially unsafe) for LBP and other health condition patients.
This project will adopt a unique approach, applying machine and deep learning to our existing large LBP patient dataset to train AI models to:
- classify LBP utilising an evidence-based classification model, enabling individualisation of prescribed exercise
- classify exercise performance, giving feedback to be adjusted accordingly to the person’s (dis)ability.
This project is funded by Wellcome Trust Population Award.