Artificial intelligence is becoming an increasingly powerful tool in medicine, helping doctors detect disease earlier and make better treatment decisions.
Researchers have now created an AI system that may predict a person’s chance of having a stroke as far as 10 years into the future using only a standard 10-second electrocardiogram. The study was carried out by scientists from Mass General Brigham and the Broad Institute and published in JACC.
Every year, millions of people around the world experience a stroke. Many survivors are left with lasting problems involving movement, speech, memory, or independence. Preventing strokes before they happen is therefore one of the biggest goals in cardiovascular medicine.
Current methods for estimating stroke risk often depend on detailed medical histories and scoring systems that are not always used in everyday practice. The new AI model was designed to make risk prediction much simpler by using information already collected during a routine heart test.
The researchers trained their deep learning model, called ECG2Stroke, using data from more than 200,000 patients. It learned to recognize subtle electrical patterns in ECG recordings that may reflect hidden changes in the heart linked to future stroke risk. Age and sex were the only additional pieces of information required.
When tested in patients from several major hospitals, ECG2Stroke predicted future strokes with accuracy comparable to well-established clinical risk tools. This suggests the technology may be practical for large numbers of patients because ECG testing is already widely available and inexpensive.
The strongest predictions involved cardioembolic stroke, which develops when a blood clot forms in the heart and travels to the brain. Because doctors can often prevent this type of stroke with blood-thinning medication, identifying these patients earlier could have a major impact on public health.
The AI system also highlighted the importance of the atria, the heart’s upper chambers. Subtle electrical changes in these chambers appeared to provide important clues about stroke risk years before an event occurred. This may also help researchers better understand the biological processes behind stroke.
Despite the promising results, the authors caution that the model still needs to be tested in prospective real-world studies before becoming part of routine care. Researchers must confirm that using the tool actually changes treatment decisions and reduces the number of strokes over time.
If successful, the technology could help doctors quickly identify patients who need closer monitoring, additional heart testing, or preventive therapies. Because the ECG is already a standard medical test, introducing the AI model may require relatively few changes to current healthcare systems.
In analysis, this research demonstrates how artificial intelligence can uncover valuable information hidden inside routine medical tests. ECG2Stroke offers the possibility of identifying stroke risk years earlier than before using a simple, low-cost examination. Continued research will determine whether this promising technology can improve prevention and save lives.
If you care about stroke, please read studies that diets high in flavonoids could help reduce stroke risk, and MIND diet could slow down cognitive decline after stroke.
For more health information, please see recent studies about antioxidants that could help reduce the risk of dementia, and tea and coffee may help lower your risk of stroke, dementia.
Source: Mass General Brigham
