An ECG biomarker for sudden cardiac death discovered with deep learning
Researchers led by UC Berkeley used a deep learning model to identify a previously unknown ECG biomarker for sudden cardiac death. The system detects a slurred downstroke after the R-peak in routine ECGs that doctors previously cleared as normal. This tool could help identify thousands more high-risk patients annually who may require internal defibrillators.
What changed
Deep learning has identified a specific physiological signal in ECGs to predict sudden cardiac arrest.
Live updates
-
AI Discovers New ECG Biomarker for Sudden Cardiac Death Risk
confidence 90%Researchers led by UC Berkeley used a deep learning model to identify a previously unknown ECG biomarker for sudden cardiac death. The system detects a slurred downstroke after the R-peak in routine ECGs that doctors previously cleared as normal. This tool could help identify thousands more high-risk patients annually who may require internal defibrillators.
What's confirmed:
- UC Berkeley researchers developed an AI model that identifies a hidden ECG signal to predict sudden cardiac death.
- The deep learning model identifies a slurred downstroke after the R-peak as a biomarker for sudden cardiac death.
- The AI system can detect risk in routine ECGs that physicians had previously marked as normal.
- More than 300,000 people die annually in the U.S. from sudden cardiac arrest.
Still unconfirmed:
- Pathway Labs launched an FDA-approved EchoNext AI to detect hidden heart disease from ECGs.
- A 75-year-old woman in Sweden was admitted to the emergency room with dizziness and a normal echocardiogram.
- Pathway Labs raised $8.5M.