AI Predicts Stroke Risk From DNA Like ChatGPT Reads Words

Scientists just taught artificial intelligence to read your DNA the same way ChatGPT reads sentences, and it could help save thousands of lives from stroke.

A team at the University of Technology Sydney developed an AI model that scans a person's genetic code to spot hidden stroke risks. Using the same transformer technology that powers ChatGPT, the tool identified 18 genetic variants linked to ischaemic stroke, the most common type where blood clots block brain blood flow.

The breakthrough matters most for families with stroke history. "When someone has a stroke, particularly at a young age, they often want to know whether other family members are also at risk," said researcher Kairui Guo. These new tools finally give doctors a way to answer that question with real precision.

Genetics account for about 38% of stroke risk, yet doctors rarely factor DNA into prevention plans. Processing that much genetic information for each person would be impossible without AI doing the heavy lifting.

The team tested their model on more than 40,000 people across two major medical datasets. It successfully sorted people into low, medium and high genetic risk groups. Those flagged as high risk were 21% more likely to suffer their first stroke compared to the low-risk group.

The AI outperforms older methods that simply count up risky genes. Traditional polygenic risk scores miss how genetic variations interact with each other or how their position in DNA affects overall risk. The new model catches those subtle patterns.

The Ripple Effect

This technology opens doors far beyond a single test result. Doctors can now combine genetic risk scores with lifestyle data like blood pressure, age, diabetes status and cholesterol levels to create complete prevention plans tailored to each person.

The Australian Artificial Intelligence Institute is already building the next version. Their multimodal AI tool will blend genetic assessments with health surveys, medical scans and imaging to paint an even clearer picture of stroke risk.

Two genes stood out in the research: EDEM2, active in the brain cortex, and KANK2, found in blood. Understanding which specific genes drive stroke risk helps researchers design targeted treatments that work with your body's unique makeup rather than against it.

The technology transforms a terrifying question into an answerable one, turning genetic destiny into a roadmap for prevention.