An AI system trained on routine heart scans identified a group with a higher reported annual risk of sudden cardiac death than conventional screening, according to a report on research by University of California, Berkeley scientists. The approach could help doctors decide who needs closer assessment, but the researchers say it is not ready to determine who should receive an implanted defibrillator.
Berkeley BAIR Watch analysis
What happened
The team trained its system on more than 440,000 electrocardiograms (ECGs) from Sweden, linking the recordings to death-certificate information. The report says the researchers then tested it on separate patient records from a San Diego hospital system and from Taipei.
In the comparison described, conventional screening identified a group with a 4.6% annual rate of sudden cardiac death; the AI identified a group with a 7% rate. The system also flagged a broader group that existing guidelines might consider relatively low risk. The SSBCrack report says the study was published in Nature and quotes lead author Ziad Obermeyer. It does not establish that the system is ready for clinical decision-making.
Why it matters
ECGs are already collected in hospitals and medical centres, so the proposed next step could be to search existing records for people who might benefit from further monitoring. That could help clinicians focus follow-up testing on patients whose risk might otherwise go unnoticed.
The comparison also points to a difficult trade-off: current assessments can miss people at risk, while many patients given implanted defibrillators never need the device to deliver a shock. A better way to identify risk could matter, but identifying a higher-risk group is not the same as proving that acting on the prediction improves outcomes.
Our read
This is a promising research result with a practical route towards further testing: use existing ECG records to identify people who may need closer assessment. The figures are striking, but the sensible next step is clinical evaluation, not handing a treatment decision to an algorithm. A scan is not a verdict, however clever the pattern-matching.
What to watch
- Whether further studies confirm performance in clinical settings and across different patient groups.
- What follow-up monitoring finds in people flagged by the system.
- How clinicians could use the risk estimates alongside existing assessments and patient needs.
Discussion spark: If an AI flags a patient as higher risk using an ECG already on file, should that be enough to prompt follow-up testing, or should the evidence bar be higher?
Sources and evidence
- AI Detects Hidden Sudden Cardiac Death Risk in Routine EKGs – SSBCrack (15 November 2026, 05:28 UTC)
Independent WittyWires tracker for public updates about Berkeley BAIR. Not affiliated with or endorsed by Berkeley BAIR; this is not an official account.