A routine heart test may hold a warning sign that doctors have missed for years. Researchers at UC Berkeley have trained an artificial intelligence model to study ECGs and look for patterns tied to sudden cardiac death. This condition can strike people with known heart problems, as well as younger athletes and individuals who never knew they were at risk.

How AI Can Help

The AI model was trained on over 440,000 ECGs from Sweden and paired with death certificates and health records. It was then tested on separate patient data from the U.S. and Taiwan. The model found a high-risk group with a 7% annual rate of sudden cardiac death, compared to the standard method which had a 4.6% annual rate.

The researchers also used another AI system to compare low-risk and high-risk ECG patterns. This comparison pointed to a visible feature in one part of the ECG called aVL, which strongly predicted sudden cardiac death. This signal had not been previously described in medical literature, raising the possibility that AI may help doctors make better predictions and spot warning signs humans have missed.

Implications and Next Steps

The next phase of the research is already underway, with health systems in Sweden, Taiwan, and the U.S. testing the algorithm on hospital ECG databases. If the tool flags a scan as high risk, doctors could contact the patient and have them wear a heart-monitoring patch to reveal more about the dangerous rhythm before it turns fatal.

While this AI tool is promising, it still needs more testing before it becomes part of routine care. Doctors need to know it works across more patients, and hospitals need a plan for what happens after an AI alert. Patients also deserve clear privacy protections when their medical scans help train these systems.

Original reporting: Fox News (HLL/CB) — read the source article.