Home Smart Home

For most people, using toilets and washing their hands are such simple tasks they barely think about them. But for some sufferers of dementia and Alzheimer’s disease, they can seem confusing and overwhelming.

“What we typically see [with Alzheimer’s patients] is they go into the washroom and just stand there, and don’t remember why they’re there,” says Alex Mihailidis, an assistant professor of occupational science and occupational therapy at U of T. Through his research in the Intelligent Environment Laboratory at the groundbreaking iDAPT (Toronto Rehab’s Intelligent Design for Adaptation, Participation and Technology facilities), Mihailidis is working on technologies that will restore a measure of independence and privacy to sufferers of dementia. iDAPT, a suite of high-tech laboratories still under construction, will be one of the world’s most advanced centres for research into disability and rehabilitation when it is completed in 2011.

Mihailidis and his colleagues have built an artificial intelligence system that can recognize when patients need help and prompt them with instructions. Cameras in the ceiling track the user’s movements and behaviour; computer software interprets what the cameras see, detects what kind of help is needed and provides spoken instructions or even video tutorials to prompt the patient to the next action. The software adapts to its user, learning what level of prompting is required. “With one person it can provide a generalized cue, where it would say, ‘Dry your hands,’” says Mihailidis, “whereas another person might require a little more detail, so the prompt may be structured to address the person by name, and play a video that shows the person how to turn the water on or how to use the soap.”

Variations on this system are already in use in three Toronto nursing homes, and Mihailidis and his team are using those real-world experiences to refine the software further. He’s also developing a similar system to detect when a person has fallen down. Today’s commercially available medical-alert systems for elderly patients require that the person remember to put on their necklace or bracelet every day, and be able to activate it if they fall; with artificial intelligence and computer vision, Mihailidis’ system could recognize any sign of trouble and call for help on its own. He is also investigating how the system could help children with autism.

Mihailidis says the ultimate goal of his research is an “intelligent home” in which artificial intelligence means “the house understands your intent, and provides assistance to you in a non-invasive fashion.” Along with dementia sufferers, it could also help with everyday forgetfulness: “It’ll recognize that I’m walking out the door without my car keys,” says Mihailidis, “and it’ll say, ‘Alex, your car keys are in the living room.’”