VCU-developed home sensor system could spot loneliness in older adults

By Madeline Reinsel

Loneliness is common in older adults, and the mental and physical health risks associated with the condition are difficult for medical professionals to recognize and treat. To address a need, researchers from Virginia Commonwealth University and Emory University have developed a passive home sensing system that could detect behavioral patterns associated with loneliness in older adults and report those results to a caregiving team.

The system, which combines commercially available and custom sensors, passively measures movement, daily activities, sleep and phone usage within the home. The team deployed the prototype system, dubbed HomePal, for three months in the homes of 70 adults ages 65 and over living independently in metro Atlanta. The research team plans to compare the data against biweekly loneliness assessments.

"The main idea is to see whether everyday behavior at home can give early signs that an older adult may be experiencing loneliness or social withdrawal," said Mohamed Kamel, a computer science Ph.D. student in VCU's College of Engineering and lead author of the study about the system. The project, co-led by VCU computer science professor Tamer Nadeem, Ph.D., and Jane Chung, Ph.D., of Emory University, is supported by a $609,933 grant from the National Institute on Aging.

How the HomePal system could identify loneliness

The HomePal system includes the following:

• Motion sensors and door sensors to detect movement inside the home, and possibly flag social withdrawal.
• Television smart plugs to estimate screen time, which is often a sedentary or solitary activity.
• Sleep sensors to track rest patterns. Poor sleep could point to symptoms of anxiety or depression.
• Smart speakers to detect loneliness-related vocal patterns.
• And phone call patterns to measure social connectedness.

"Through your regular daily activities, we can actually collect, monitor and raise a flag either to the patients themselves or to their caregiver," Nadeem said.

How HomePal data, surveys and AI intersect

The researchers' goal is to compare daily living patterns with reported feelings of loneliness, collected for each participant via assessments led by Chung, a former VCU School of Nursing professor.

The VCU team is currently using an AI model to determine which sensed behavioral patterns are more strongly associated with loneliness, and how those patterns match up with loneliness assessment results. That could help caregivers track behavior over time and flag worrying changes.

"The idea is that we are collecting these data, and then we are analyzing these data and building our models," Nadeem said. "And then we will verify how accurate our models are compared to the traditional survey approach."

Similarities to a home security system

Ultimately, the researchers hope to create a commercial product that can be installed throughout any home, similar to a home security system.

The tracking system would produce a daily loneliness score or trend, which could then be used by family members or medical caregivers to guide care and treatment.

"Loneliness is hard to detect in a clinic visit, because older adults often don't report it," Chung said. "Bringing together nursing and computer science lets us connect what we see in everyday behavior with what people tell us about how they feel, and that's what could make early intervention possible."

A focus on data and privacy

HomePal, which currently uses smart speakers and smart television plugs, could also be modified to only use sensors that operate and store data on local networks, increasing privacy.

"We have this flexibility to decide which data could be shared with the caregiver," Nadeem said. "And the data itself would be either stored locally or could be stored so that no one has access to it except the patient themselves."

But for now, the researchers plan to keep analyzing data, developing AI models and adjusting HomePal's sensors as they discover new insights.

"We are not going to say we are diagnosing loneliness directly," said Kamel, the Ph.D. student. "We are just trying to understand the patterns that may be related to loneliness and may deserve a follow-up."

Research reported in this story was supported by the National Institute on Aging of the National Institutes of Health under award number R21AG083414. One-hundred percent of the funding was from federal funds. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.