Home sensors could flag dementia earlier than doctors

Researchers in Singapore are testing a system called SINEW — Sensors In-Home for Elder Well-being — that tracks daily behaviors of older adults through motion detectors, door sensors, sleep monitors, and wearable devices. The goal: to spot the subtle changes that often precede dementia before they show up on a doctor’s exam.

The project, which aims to follow 200 seniors over three years, has released early findings from 87 participants, average age 76. Compared to peers with no cognitive issues, people with mild cognitive impairment (MCI) were less active, had poorer sleep, and were more likely to forget items around the home. Standard screening exams like the Mini-Mental State Exam missed these differences. Machine learning models trained on the sensor data, by contrast, reached strong accuracy, with one method scoring 0.91 across precision, recall, and F1-score.

“Implementing readily deployable sensor systems within community settings holds the potential for delaying or reversing disease progression,” the researchers wrote in the study, published in the International Journal of Neuropsychopharmacology.

Digital phenotyping takes hold

The SINEW project is part of a growing push toward “digital phenotyping” — using passively collected data from everyday devices to map health and behavior. In psychiatry, smartphone sensors have already been used to track sleep and cognition in people with schizophrenia, with encouraging results. In dementia research, participant-driven brain health platforms have shown older adults are able and willing to engage with wearables and apps when the systems are tailored to their needs.

The common thread is an effort to move beyond sporadic clinic visits toward continuous monitoring, capturing health changes as they unfold in real time. For dementia, that shift could mean identifying cognitive decline at a stage where interventions — lifestyle adjustments, medications under development, or clinical trial enrollment — might have more impact.

Promise and pitfalls

The appeal is obvious. In the U.S., more than 6 million people are living with Alzheimer’s disease, and the number is expected to double by 2060, according to data from the CDC. Earlier detection has become a central goal of both research and policy, but brain imaging and spinal taps are costly and invasive. Home-based digital monitoring could offer a scalable, lower-cost alternative.

But the approach raises thorny questions. Who owns the data generated by motion detectors and bed sensors in private homes? How should it be stored, shared, and secured? Dementia patients are especially vulnerable, and researchers warn that strong ethical guardrails are essential. As one commentary in Dementia Researcher noted, digital phenotyping requires “careful governance to avoid misuse”.

From study to scale

The SINEW team isn’t alone in pursuing these ideas. Other groups are experimenting with robot-assisted living environments, federated AI systems for Alzheimer’s biomarkers, and remote agitation detection tools. Together, they represent a wave of innovation trying to make dementia care more proactive rather than reactive.

Still, translating pilot studies into standard practice is a challenge. Clinicians will need evidence that digital monitoring improves outcomes, not just detection. Health systems will need reimbursement models that cover sensor installations and data analysis. And patients and families will need reassurance that such monitoring enhances, rather than intrudes upon, daily life.

A cautious optimism

For now, the SINEW study offers an early glimpse of what continuous, passive monitoring could contribute to dementia care. If the results hold up in larger cohorts, they could pave the way for a new class of diagnostic support tools — less about high-tech brain scans, more about the rhythms of walking, sleeping, and moving through a home.