AI Model Detects Depression in Older Adults Using Wi-Fi Motion Data with 87% Accuracy

Traditional depression detection methods can be intrusive and inaccessible for older adults. Now, researchers have developed HOPE, an AI model that passively monitors movement and sleep patterns via Wi-Fi signals—spotting early signs of depression with remarkable accuracy, without wearables or active user participation.

Research: Development and Feasibility Study of HOPE Model for Prediction of Depression Among Older Adults Using Wi-Fi-based Motion Sensor Data: Machine Learning Study. Image Credit: Gorgev / Shutterstock

A new study published in JMIR Aging developed and tested an innovative artificial intelligence (AI) model called HOPE that uses Wi-Fi-based motion sensor data to detect depression in older adults. This study offers a nonintrusive alternative to wearable devices, improving accessibility and compliance among aging populations. The research, titled "Development and Feasibility Study of HOPE Model for Prediction of Depression Among Older Adults Using Wi-Fi-based Motion Sensor Data: Machine Learning Study," was published by JMIR Publications. It highlights a novel machine learning model that accurately detected depression among older adults.

Led by Professor Samira A Rhaimi from McGill University and Mila-Quebec AI Institute as principal investigator, the study aimed to determine whether everyday movement and sleep patterns collected through Wi–Fi–based sensors could provide early indicators of depression in adults 65 years and older. With an accuracy rate above 87%, this innovative approach presents a promising solution for early intervention and nonintrusive mental health monitoring, offering an alternative to traditional methods that require direct patient engagement.

Structure of the automatic Wi-Fi–based depression classification framework. CSI: channel state information; DT: decision tree; EFS: Edmonton Frailty Scale; GDS: Geriatric Depression Scale; LIME: local interpretable model-agnostic explanations; PCA: principal component analysis; RSSI: received signal strength indicator; SFS: sequential forward selection; SHAP: Shapely addictive explanations.

Depression is a growing public health concern among older adults, with studies estimating that 10%-15% of community-dwelling older adults and 30%-40% of those in long-term care facilities experience this condition. However, nearly half of depression cases remain undiagnosed, leading to detrimental effects on physical health, increased hospitalization rates, and reduced quality of life. Traditional detection methods, including clinical interviews and wearable-based monitoring, are often resource-intensive, intrusive, or inconvenient, particularly for older adults who may struggle with technology adoption. The HOPE model addresses these challenges by leveraging existing Wi-Fi infrastructure, enabling continuous passive monitoring without requiring any active participation from users.

A key aspect of the HOPE model is the integration of explainable AI (XAI) techniques, ensuring transparency and clinical interpretability. Rahimi's lab used explainable machine learning models to identify the most influential factors in depression detection. The results underscored the important role of sleep-related features, including average sleep duration, frequency of sleep interruptions, and frailty levels as primary indicators of depression. By making these AI-driven predictions interpretable and clinically meaningful, the HOPE model enhances trust and facilitates early detection of depression among older adults in the community.

The study highlights the importance of sleep-related factors in detecting depression. The analysis revealed that the most influential factors were sleep duration, the number and duration of sleep interruptions, and the level of frailty, which aligns with previous research on the link between sleep and mental health and reinforces the need for further exploration in this area.

"Too often, the mental health of older adults is overlooked, leaving many to suffer in silence without the care and attention they deserve. Our HOPE model could act as a caring friend who looks out for signs of depression in older adults using everyday Wi-Fi data to spot potential issues early on and without being intrusive. It's about using technology to lend a helping hand, especially for those who might find it hard to reach out themselves," said Samira A. Rahimi, one of the McGill University researchers.

The study demonstrates the feasibility of using smart home technology for mental health assessments. While these findings are promising, more extensive studies are needed to provide further evidence for this approach. This technology could support early intervention efforts and improve the quality of life for older adults at risk of depression.