Dr Kolanko said the tech can ‘act as an early warning tool’ (Image: LifeArc)
A bed sensor that uses artificial intelligence to spot early signs of infection in people with dementia could prevent severe illnesses and slash hospital admissions, according to researchers. The system uses a contactless sensor placed under a mattress, which monitors changes in heart rate and breathing patterns overnight. Readings are then analysed by AI to identify unusual patterns and alert caregivers if red flags are spotted.
Infections are one of the leading reasons people with dementia are admitted to hospital, but patients may struggle to communicate symptoms or notice changes. Study leader Dr Magdalena Kolanko, an NHS doctor and a clinical research fellow at the UK Dementia Research Institute, said: “This study is important because it shows that health changes linked to infection may be picked up passively at home, without asking people living with dementia or frail older adults to wear or manage a device.
“That matters because wearable technology can be uncomfortable, easy to forget, or hard to use over long periods. This technology is cheap, and therefore easily scalable.
“In the future, this approach could help care teams notice possible illness earlier, decide when a closer check is needed, and support faster treatment while someone is still at home.”
The technology was tested through MinderCare, a home monitoring service run by Imperial College Healthcare NHS Trust. Some 180 people living with dementia contributed more than 25,783 nights of monitoring data.
The system first learned what usual physiological patterns looked like for each user, before looking for any changes away from that individual baseline.
If a user’s infection risk score rose above a set threshold, the system sent an alert to their monitoring team.
Around one in three high-risk alerts were confirmed as acute infections, and half of the alerts were linked to a clinically relevant health event. Researchers said this was a good result for this type of screening tool.
The AI correctly identified around six in ten people who had a respiratory infection, while correctly ruling out infection in almost nine out of ten people who were well.
Under-mattress and room-based sensors are already available for care monitoring, however the researchers aimed to develop a more advanced approach to making clinical predictions than many commercial systems offer.
Dr Kolanko said the technology would not replace a doctor, nurse or standard medical tests but could “act as an early warning tool”.
She added: “It may help care teams decide when someone needs a closer check, while they are still at home, preventing the severe consequences of undetected infections, such as increased confusion, falls, and an escalation of care that leads to avoidable hospitalisations.”
The sensor is placed under a mattress to take readings overnight (Image: LifeArc)
Dementia costs the NHS around £2.7bn per year, with emergency admissions accounting for around 90% of the total. Almost two thirds of emergency admissions for people with dementia are for avoidable illnesses.
The research was funded and supported by the medical research charity LifeArc, in partnership with the UK Dementia Research Institute.
Sam Barrell, LifeArc’s chief executive, said minor illnesses caused by infections can quickly escalate and “lead to hospital admission, a loss of independence, and significant distress for families”.
She added: “Technology that can detect subtle changes before symptoms become obvious could give families and healthcare professionals a valuable opportunity to intervene earlier.
“A hospital admission can be a major setback for someone with dementia. If we can spot problems earlier and help people stay safely at home, that has the potential to change lives, not just ease pressure on health services.”
The findings will be presented at the world’s largest annual gathering of dementia scientists, the Alzheimer’s Association International Conference, in London this week.
‘The impact of this simple device was astonishing’
Manuel Melim, 87, developed early-onset dementia after experiencing a brain bleed in 2024. He often became confused and misplaced things around his home, and occasionally saw things that were not there.
He was supported by his daughter and granddaughter, with help from MinderCare. Manuel said: “We were then given the opportunity to use an under‑bed sensor. At the beginning, we were somewhat sceptical.
“We underestimated the ability of this small, discreet device sitting underneath my mattress. Over time, it proved to be astonishing. The sensor was able to detect changes in my health and alert the team, who in turn notified my granddaughter.”
The system picked up on early signs that Manuel was developing an infection, feeling unwell or experiencing pain on numerous occasions. This meant he could receive treatment sooner, before his symptoms escalated.
Manuel, of north-west London, added: “Over the past two years, this system has helped me tremendously and has even prevented several hospital visits, as issues were identified early.
“This technology has given all of us great reassurance and peace of mind. It feels like an extra layer of support watching over me while I sleep.
“We are incredibly grateful to the researchers and organisations involved, including LifeArc and the UK Dementia Research Institute.
“This under‑mattress sensor has shown us that it has real potential to make a meaningful difference for people like me, even in the early stages of dementia, and for the family members who care for us.”
