Can AI Predict Who Will Get Dementia?
A new AI tool to predict dementia is being developed by the University of Auckland and a team of experts from Singapore.

The Ministry for Business, Innovation and Employment has just granted $4 million for the project led by Waipapa Taumata Rau senior research fellow in psychology Dr Catherine Morgan, psychology Professor Lynette Tippett, senior lecturer in computer science Dr Katerina Taskova, and University of Otago neuroscientist Dr Narun Pat.
At present, health professionals can struggle to assess whether a person is likely to get dementia, based on reports of their memory loss and ability to function from patients and families.

The team aims to create an AI tool that health professionals can use to more accurately predict who is at risk of developing dementia.
Artificial intelligence (AI) can easily combine large amounts of data that can be used to predict dementia, such as information about memory test results, brain imaging, blood tests, genetics, vascular health, and general health, says Morgan.
“We’re aiming for more than 85 percent accuracy in predicting dementia risk.
Advertisement - scroll to continue reading“Our ultimate goal is to be able to pick up the early warning signs, then we might be able to do something to delay or prevent the progression of dementia,” says Morgan.
AI will allow the team to understand which factors contribute most to dementia risk, and which combinations of factors pose the greatest threat. This will offer new scientific insights that could inform the development of future treatments and tests for the disease.
“We’re excited because this could identify new areas for developing pharmaceutical treatments.
“We might also find there are additional tests people should have that would catch dementia early,” says Morgan.
Research has uncovered lifestyle factors that can help keep dementia at bay and new treatments are in the pipeline, says Tippett.
“We want to find a way to identify people who are at highest risk of dementia, before it develops. There might be a chance to intervene before too much damage is done to the brain,” Tippett says.
People with high risk could be sent for further tests with a specialist and provided with information about what to expect.
“Being able to understand what’s going on and make life plans can make a huge difference for families,” says Tippett.
The tool will provide medical professionals with a risk score, indicating the likelihood of a person developing dementia. It will offer explainable AI, so health professionals can see which factors the tool has picked up and how the risk score was calculated.
The project is driven by a “massive problem”, which is growing in New Zealand and globally as people live longer and the population ages, says Morgan.
“About 70,000 people are affected by dementia in New Zealand at the moment, but that is projected to more than double by 2050,” Morgan says.
The healthcare cost of dementia is about $274 million a year in New Zealand, but is projected to rise to $658 million by 2050.
Dementia can be difficult to identify, particularly among people aged under 65, who make up about nine percent of dementia cases, says Tippett.
Predicting dementia is particularly challenging, because some people with mild cognitive impairment rapidly decline into Alzheimer’s and dementia, while others remain stable, says Morgan.
Magnetic resonance imaging (MRI) scans of the brain and markers in blood often show the earliest changes, before dementia symptoms appear.
This and other data has been collected in long-term health studies in New Zealand and Singapore, which will be used to create the software to predict dementia.
“We have collected an incredible wealth of information through our Dementia Prevention Research Clinics over the past nine years and that has put us in a position to be able to help develop this tool.
“We know which patients developed Alzheimer’s and dementia and who stayed stable, so we can train the AI algorithm using that,” says Tippett, who is Director of the Dementia Prevention Research Clinics.
Other studies providing data for the project include the New Zealand Brain Research Institute’s Parkinson’s Progression Programme, the Dunedin Multidisciplinary Health and Development Study, and similar programmes in Singapore.
The academic team is partnering with Siemens Healthcare to develop brain-image analysis tools that will be used at the time of MRI scanning for faster transfer of scan results.
They are co-designing the tool with doctors, to make sure it’s user-friendly.
In three years, the team hopes to have developed the AI tool and to have tested it in a pilot study.
The project team includes University of Auckland’s Centre for Brain Research deputy-director and Māori adviser Dr Makarena Dudley, dementia specialists from Te Whatu Ora, and National University of Singapore Associate Professor Saima Halal and her team.
The research builds on a $600,000 Health Research Council Sir Charles Hercus research fellowship provided to Morgan last year to study magnetic resonance imaging (MRI) to predict dementia. Brain Research New Zealand, the New Zealand Dementia Prevention Trust, Hugh Green Foundation, and the Angus Trust have also made grants paving the way for this project to go ahead.