Battling ageism and ablism in AI to prevent dementia patients from falling

Dr. Babak Taati has been named a 2020 AMS Compassion and AI fellow. During his fellowship, his goal is to apply a holistic, user-centred approach to improve efficacy and address underlying issues.

Approximately 60 per cent of older adults with dementia fall each year. With such a troubling statistic, it’s no surprise that falls are a leading cause of injury, loss of independence and mortality in older adults with dementia.

To address this growing problem, scientists like Dr. Babak Taati at the Kite Research Institute are dedicating their entire research focus to understanding, designing and developing technologies that can predict and prevent this vulnerable population from falling to the ground and suffering injury. 

Taati believes his recent appointed as a AMS Compassion and AI fellow brings him one step closer to this goal. The fellowship will allow Taati to develop intelligent systems that serve vulnerable populations while eliminating ageism and ableism in artificial intelligence (AI) technology.  

While falls prevention systems that use AI technology already exist – including those with facial recognition capabilities – Taati says they don’t work well to evaluate the faces of older adults with dementia.

“Facial recognition technologies work well for young, healthy adults, but not as well for older adults with dementia, cerebral palsy and Parkinson’s.” says Taati. 

Taati’s first step will be to develop and validate a holistic machine learning model that combines patterns of gait and available medical information to create a short-term fall prediction system. This is especially important because fall risk in dementia is dynamic and can fluctuate. 

There are multiple factors that increase fall risk, such as taking new medications, weakening health and declines in cognition. While preliminary results with classical machine learning models are promising and can predict a fall within 30 days, he is looking to improve these results further. 

He also plans to phase out the use of RFID tags in the AMBIENT system, which was developed right here at KITE, and replace it with automated facial recognition technology.

Lastly, he intends to focus on applying learning and privacy-preserving machine learning techniques to reduce biases against older adults with dementia and to improve system performance. 

Through compassionate, inclusive technology, Taati’s research aims to improve quality of life by identifying and preventing injury effectively. This is, and will be, a huge leap forward in the world of machine learning, facial recognition and other sophisticated healthcare technologies.