Waterloo researchers test AI for monitoring seniors
April 12, 2023
WATERLOO, Ont. – Researchers at the University of Waterloo in Ontario are using AI and non-invasive technology to monitor seniors in long-term care facilities – reducing the need for cameras, fobs or other traditional wearable gadgets. Instead, the research team is testing a wall-hung, low-power radio system and artificial intelligence (AI) to take note of habits like how often residents go to the washroom, when they eat, or how long they usually watch TV. It can also alert care providers in the event of a fall.
Hajar Abedi (pictured), a PhD candidate in systems design engineering at the University of Waterloo, led a study into using AI to monitor seniors in long-term care without the need for wearable devices or cameras. The study was published in the journal the Institute of Electrical and Electronics Engineers (IEEE) Internet of Things. “We use artificial intelligence to actually make our lives easier because we can train them and they can do our job, and basically, our main purpose is to save lives using this AI technology,” Abedi told CBC Kitchener-Waterloo.
The monitoring system, called ElephasCare, works via a radio wave device similar in size to an internet modem and can be mounted on the wall. No cameras are installed.
Monitoring seniors as they go about their daily lives allows care providers to keep on top of any changes, she said. “For example, we know your gait speed right now and in the next month. So based on what we see, we have a baseline of each individual,” Abedi said.
When the monitor notes something abnormal, such as a recurring change in stride length while walking, “we can identify or send notification to caregivers. If [the seniors] walk slower over a period of time, [caregivers] can tell.”
Without having to use cameras, this helps protect a person’s privacy and is more beneficial than wearable devices, which can be cumbersome or sometimes require the person to interact with them to ensure they’re working properly, Abedi said.