Researchers at the University of Technology Sydney (UTS) have developed an AI model that uses genomic data to predict ischaemic stroke risk, identifying 18 genetic variants associated with the condition and outperforming existing genetic risk prediction tools.
The model was tested using genetic and health data from the UK Biobank and the US-based All of Us research database, with researchers finding that people in the highest genetic risk group had a 22 per cent higher chance of stroke than those in the lower risk group.
The study, Predicting risk of ischaemic stroke: A transformer model using genomic data, used transformer-based AI technology to analyse thousands of genetic variations, taking into account their location in the genome and potential interactions.
The researchers also identified the EDEM2 gene in brain cortex tissue as a key contributor to stroke susceptibility, providing a potential target for further investigation.
UTS researcher Dr Yue (Eric) Yang, who recently completed a PhD in artificial intelligence and was first author of the study, said the model was more accurate than previous genetic risk tools at distinguishing people who had experienced a stroke from those who had not.
“Genetic susceptibility plays an important role in stroke risk, along with lifestyle factors such as smoking and alcohol consumption,” Dr Yang said.
“We tested our AI model using genetic and health data from the UK Biobank, and also against health records from a large US-based database (All-of-Us). Our model was more accurate than previous genetic risk tools at distinguishing people who had suffered a stroke from those who had not.”
UTS Australian Artificial Intelligence Institute director Distinguished Professor Jie Lu said combining genetic information with established clinical risk factors could improve personalised stroke prevention.
“AI can be used to find associations between human genes and disease, and build personalised risk assessments, to help prevent a potentially life-changing or fatal event,” Professor Lu said.
“Genetics are only one part of stroke risk. The ability to combine a genetic score with clinical and lifestyle information, including established risk factors such as blood pressure, age, atrial fibrillation, type 2 diabetes and cholesterol, could enhance monitoring and prevention.”
The researchers are now developing a multimodal AI model that would combine genomic risk assessments with health and lifestyle surveys, scans and imaging to improve risk prediction for stroke and other conditions, including endometriosis.
Corresponding author Dr Kairui Guo has also been working with the stroke unit at Liverpool Hospital in Sydney on genomic analysis and AI applications for acute stroke management, supported by neurologist Associate Professor Dennis Cordato.
Dr Guo said the team was proposing a preventative clinical trial involving genome sequencing and stroke risk assessment, with plans to investigate differences in genetic risk between men and women.
“When someone has a stroke, particularly at a young age, they often want to know whether other family members are also at risk. These new AI tools will provide the ability to better help families understand that risk,” Dr Guo said.
“We currently have a proposal for a preventative clinical trial. The aim is to recruit the first cohort of patients, sequence their genome, and analyse stroke risk.”
He said genetics had traditionally received less attention in stroke risk assessment, despite its potential contribution to identifying people at greater risk.
“Without AI, it would be impossible to process that much data at an individual level to provide a personalised assessment.”
Add Comment