Studies under way to use AI to identify seniors at risk for depression

AI model, named HOPE, uses WiFi sensing to analyze seniors’ daily movements at home

A researcher in the Lady Davis Institute (LDI) at the JGH is laying the groundwork for what she hopes will be the eventual use of artificial intelligence to help predict the risk of depression in seniors.

In a study published in JMIR Aging, Professor Samira Abbasgholizadeh Rahimi, a Principal Investigator at the LDI, used AI and WiFi sensing technology to determine that depression is likelier to occur in older individuals whose sleep is regularly interrupted and/or insufficient.

After additional research has been conducted, Prof. Rahimi says the findings could become the basis of a smart tool to help caregivers, family doctors, geriatricians and other healthcare professionals to spot the warning signs of depression in the elderly.

“Certain healthcare professionals are already trained to be alert for potential problems in their older patients,” says Prof. Rahimi, Co-Director of the McGill Collaborative for AI and Society. “My hope is that someday AI can complement their skills by noting the possibility of depression at an earlier stage.”

Prof. Rahimi, an Assistant Professor at McGill University and at the Mila-Quebec AI Institute, developed the Home-based Older Adults’ Depression Prediction (HOPE), a new AI model that monitored the everyday activities and sleep routines of five seniors (65 years old and up) for six months in 2023.

Collecting data non-intrusively

This was accomplished by installing two or three smart sensing WiFi devices in each residence—one in the bedroom, another in the living room and sometimes, if the home was large, a third elsewhere. As a result, information about the test subjects was collected non-intrusively and with a high degree of accuracy.

The data was transmitted to Prof. Rahimi’s lab, where it was analyzed by the HOPE AI model, which monitored time spent sleeping, preparing meals, moving between rooms and performing other activities. This allowed HOPE to detect behavioural patterns associated with early signs of depression.

Wi-Fi sensing was used, because Prof. Rahimi knew from experience that many elderly people are put off by the intrusiveness of digital wearables or similar devices that are in contact with the body for extended periods.

“We wanted to use data-collection technology that would be uninterrupted, accurate and not burdensome to these individuals,” she explains.

In addition to collecting demographic data about the seniors, Prof. Rahimi’s team administered a set of clinically validated assessments that are commonly used to evaluate depression, physical frailty and cognitive function in older adults.

To add a personal element to the study, she and her colleagues also spoke with the seniors to gain a clearer understanding of what they were feeling and experiencing during their day-to-day routines.

In developing the HOPE model, the team analyzed extensive Wi-Fi data alongside clinically assessed outcomes. Among several key findings, the model identified a clear association between depressive symptoms and patterns of disrupted or insufficient sleep in seniors.

Although the AI program took various types of home-based activities into consideration, sleep was singled out as the factor deserving the most attention, Prof. Rahimi adds.

Expanding the research

“It’s exciting to see researchers and other professionals exploring the potential of AI in so many different ways,” says Professor Maxime Cohen, Chief of AI Strategy for CIUSSS West-Central Montreal.

“While not every idea may succeed, AI is still having a profound impact: It’s providing renewed motivation for healthcare professionals to investigate innovative, new approaches, in order to improve the quality, accuracy and effectiveness of care.
 
“Equally important is the powerful promise AI holds as a predictive tool to help us anticipate challenges before they arise and to detect small issues before they escalate into major crises.”

Now that the basic principles have been established, the next step for Prof. Rahimi is to continue her research with a larger group of older adults.

At some point, she hopes formal and informal care providers will be able to use the HOPE model to predict the early warning signs of depression in senior patients by placing Wi-Fi devices in their homes and gathering data about their behaviour, particularly the quality of their sleep.

Prof. Rahimi explains that a tool of this kind would be ideal for elderly individuals who, for whatever reason, are reluctant to visit a clinic or have a regular medical checkup.

“It would enable their caregivers and doctors to provide them with personalized interventions that are suited to their particular needs,” she says.

“By streamlining data collection and analysis, AI allows us to spot warning signs sooner, so we can step in faster when an older adult is at risk of depression.”