Written Evidence Submitted by
UK Dementia Research Institute
(GAI0111)
The use of artificial intelligence to transform dementia care and clinical trials
The demographic challenges associated with dementia, and the attached economic costs, are stark. One in three people born today will develop dementia in their lifetime. Dementia is consistently the leading cause of death in England and Wales, revealing itself as our deadliest silent pandemic. Meanwhile, an estimated 900,000 people in the UK are living with dementia, and, with our ageing population, this number is predicted to reach 1.6m by 2040. In the NHS, 1 in 4 hospital beds is currently occupied by somebody with dementia. The cost of dementia to the UK economy is estimated at £34.7bn per annum, equivalent to nearly a quarter of the NHS budget.
Despite the significant economic and personal costs of dementia, there is a striking lack of therapeutic interventions. With cases of dementia expected to triple globally by 2050, our greatest healthcare challenge is growing, and the need for action is urgent. Without new breakthroughs, dementia will cost the UK economy an estimated £94.1bn per annum by 2040.
About the UK Dementia Research Institute
The UK Dementia Research Institute (UK DRI) is the UK’s national institute dedicated to stopping dementia in its tracks. Our mission is to fill the knowledge gap in dementia, and transform that knowledge into real-world tools, diagnostics and treatments. In the process, we’re marking out the UK as a world leader in dementia research, which will attract further investment.
The UK DRI was founded in 2017 by the UK Government, Alzheimer’s Society and Alzheimer’s Research UK. Our unique structure draws together the expertise of over 750 researchers at six of the UK’s top universities, and the diverse skills in their research teams.
The UK DRI is a multidisciplinary institute: 40% of our Group Leaders do not come from a “traditional” dementia background. This new approach has accelerated the pace of discoveries. We are also an international community, with one third of our Group Leaders having relocated from overseas to take up their positions.
The “Minder” platform: Harnessing AI and machine learning to transform care for people living with dementia
Too often, people with dementia are isolated and develop preventable health problems, such as infections, which can lead to unplanned hospital admissions if untreated. Today, 1 in 4 NHS hospital beds is occupied by a person living with dementia. These hospitalisations can be unsettling and, in turn, precipitate a deterioration of dementia symptoms.
The UK DRI’s Care Research & Technology Centre, based at Imperial College London and the University of Surrey, is developing the Minder platform, which uses AI and machine learning to reduce preventable hospitalisations and enable people with dementia to live independently at home. A diverse team of scientists, data experts, doctors and engineers, led by Professor David Sharp, is harnessing advances in artificial intelligence, engineering, robotics and sleep science, to build a platform that delivers the highest quality care to people with dementia in their own homes. The team is guided by people with dementia and their caregivers, and investigates ways to keep people independent, improve their general health and sleep, and reduce confusion and agitation. The goal is to empower people with dementia and their caregivers, by creating dementia-friendly ‘Healthy Homes’: intelligent environments that transform and personalise care.
The Minder platform transforms data from a wide range of low-cost sensors and devices into a powerful tool for clinicians and caregivers. Using sophisticated AI and machine learning, Minder synthesises data in real time from sensors around the home, such as sleep mats or vital sign monitoring devices, to learn each participant’s unique patterns and detect possible problems.
When a warning sign is identified, this is conveyed to an on-call monitoring team embedded within the NHS, to facilitate early interventions. The platform is currently being piloted in circa 80 private homes in the UK, and there is already evidence that hospitalisations have been reduced.
The Minder platform (image credit UK DRI)
Possible future applications of AI in dementia
Professor Sharp and his team are exploring the possibility of using Minder as a companion to run future clinical trials in dementia. The Minder platform collects fine, granular data about people’s physiology, activity and sleep, and aids participants to complete a clinically validated questionnaire to measure their cognitive and functional abilities. Professor Payam Barnaghi, programme lead for healthy home and machine intelligence at the UK DRI Care Research & Technology Centre, says this detailed, continuous data on each participant’s wellbeing could also be used to measure the efficacy and/or side effects of a new drug much more effectively and efficiently than currently available methods.
In the future, Barnaghi says it may also be possible to use longitudinal monitoring data such as sleep data, in combination with information from clinical healthcare records, to identify members of the population who may be at higher risk of neurodegeneration. Using AI, it may be possible to develop a precursor for screening, by combining data on changes in sleep and activity patterns with clinical information, to identify people who would be suitable for further tests.
Opportunities and challenges
The advancement of AI and machine learning presents exciting opportunities to transform the way we provide care and conduct clinical trials, not just in dementia but across the whole health and care system. In the NHS, we already collect large volumes of data as routine, which are not then synthesised or analysed effectively. With the development of the right tools, those data could be harnessed to inform clinicians’ decisions on everything from diagnosis to triage.
With the exception of the UK DRI, there are few research organisations with a long-term focus on developing AI for health and care. Most of this research is funded by short-term grants, but specialists in the field of AI are often attracted away from health research by high salaries in the private sector. In order for the field to advance quickly, the right research environment will be necessary. This should be a focused initiative, with a multi-disciplinary team and relatively long- term funding stability, to provide the focus and continuity to enable breakthroughs.
According to Barnaghi, there is currently no standardised way of validating an AI-based health tool or platform for healthcare applications. The medical certification process does not currently interrogate modelling decisions sufficiently, for example by asking how decisions on sensitivity or specificity were reached. To ensure a tool is robust enough for use in clinical practice, these elements must be considered. There will be no “one size fits all” solution to this, but a checklist of points to consider, validate and justify would be beneficial.
At the UK DRI we believe AI-based tools to inform clinical decision making could improve efficiency in the health system. However, the way in which these data are communicated to patients must be carefully considered. Data on risk and chance may be difficult for a lay person to understand, and clinicians should be trained to communicate and interpret them appropriately.
ENDS
(November 2022)