Machine learning aims to transform falls prediction in aged care

Very few people working in aged care need a reminder of how devastating falls can be. Overall, falls are Australia’s leading cause of injury hospitalisation and death, with the likelihood of both increasing considerably as we age.

Data shows that 60% of fall-related hospitalisations were for people aged 65 and over. Meanwhile, 94% of deaths occurred in the 65+ population. These are just some figures that scrape the surface of the problem at hand. 

Researchers at the University of Canberra are attempting to help reduce the risk of falls occurring in residential aged care settings with Electronics engineer and PdD candidate, Abishek Shrestha, undertaking a falls prediction project. 

He’s working under the supervision of Dr Maryam Ghahramani, Senior Lecturer and Program Director in Engineering at the Faculty of Science and Technology and a member of the UC Human Centred Technology Research Centre.

“I’ve been doing research in the field of human motion, motor function and mobility assessment in older people to look at their risk of falling for years. Roughly 30% of people above 65 fall each year,” Maryam explained to Hello Leaders. 

“We all know the consequences of falls in older people. It can easily turn into a hip fracture and it can also turn into death. It happened to two of my grandparents. My grandma fell down, had a hip fracture and then there was the surgery. Six to seven months after surgery she unfortunately passed away. 

“Assessing balance, mobility and motion in older people is important. We don’t want to wait until the fall happens to help people.”

Two additional co-supervisors, Professor Damith Herath (robotics) and Clinical Assistant Professor Dr Angie Fearon (physiotherapy) are also involved. The entire team has a shared belief that technology can play a far greater role in fall assessments, helping to reduce the potential for missed details during purely functional and visual assessments.

“We look at aged care from several different robotics perspectives. We are working with other colleagues on bringing robots into aged care spaces and to understand the limitations and where we can improve,” Damith said.

“Ian’s work is quite important to understand where the capabilities are currently and what technologies we can bring into facilities. It’s a really exciting time for us.”

As part of his research, Abishek is using machine learning to explore the Balance Mat system’s ability to identify clinically meaningful balance deficits. Balance Mat is a unique invention designed by Ian Bergman to accurately detect balance impairments against gold standard balance tests such as the Berg Balance Scale, Timed Up and Go test and timed single leg stance.

It still supports functional and visual assessments, all while providing performance data and scores based on that data. 

“The way people are assessed and scored is completely visual. A practitioner will look at how you’re standing and score you between 0 to 4. Practitioners have the training and they know what they’re doing, but we still have limited capacity,” Maryam said.

“Technology always helps us, but we can do more in terms of having technology that can assess someone’s balance. That technology needs to be accurate enough, valid with what it measures and also easy to work with.”

Ongoing research centred around Abishek’s PhD has seen the Balance Mat paired with a rig connected to robots – provided by Dimeth’s department – to analyse various postures and postural sways. 

This has allowed them to determine how accurate its measurements are, and whether it will translate to use for older people. 

It has also been tested with younger adults, however, the next phase of research will involve older people with and without histories of falls. About 100 participants will undergo an assessment with sway metrics recorded. 

The ultimate aim of the study is to provide evidence for community-based screening of undiagnosed balance decline among older people to activate supportive interventions earlier and prevent injury from falls.

For Maryam, another aim is to introduce a tool that assists with continuous balance monitoring in aged care, rather than just tests at the start and end of programs designed to improve balance issues. 

“If they want to see what the effect of an exercise program is they need to have a device that can continuously, on a daily basis, record someone’s balance,” she added.

As for Ian himself, who started testing out the earliest iterations of his Balance Mat in aged care about eight years ago, there is plenty of confidence that this tool can break new ground in falls prevention. 

“By my understanding, there are no devices in aged care that can actually identify people who are at falls risk so they’re left floundering a little bit. I believe we have a device that can go into aged care residences because it’s easy to use, it’s simple and it’s accurate,” Ian said.