A SMARTWATCH-BASED machine learning algorithm has demonstrated high specificity (99.99%) and moderate sensitivity (67.23%) in detecting sudden loss of pulse, suggesting potential for early cardiac arrest detection and automated emergency response.
Out-of-hospital cardiac arrest (OHCA) remains a leading cause of sudden death, with survival largely dependent on rapid recognition and intervention. Many OHCA cases go unwitnessed, delaying crucial medical assistance. Google Research developed an algorithm designed to detect pulseless events using photoplethysmography (PPG) and motion data from a smartwatch. If validated further, this technology could enable automatic emergency calls for users experiencing cardiac arrest, potentially improving survival rates.
The study, published in Nature, evaluated the algorithm across six different cohorts, including clinical and real-world settings. In a controlled electrophysiology lab, 100 patients undergoing defibrillator testing provided data on ventricular fibrillation-induced pulselessness, while another 99 subjects experienced pulselessness via a tourniquet-induced arterial occlusion model. Additionally, 948 participants contributed free-living data without pulseless events. The system’s sensitivity was 72% for motionless pulseless events and 53% for simulated collapses. It identified pulselessness within 57 seconds and initiated a 20-second user response check before calling emergency services. False alarms were minimal, with one false call occurring per 21.67 user-years.
This research highlights the potential for wearable technology to enhance early cardiac arrest detection, particularly for unwitnessed events. While the algorithm’s high specificity is promising, improving sensitivity and minimizing false positives remain critical before widespread deployment. Further real-world testing and refinement are necessary to ensure its effectiveness across diverse conditions. If successfully implemented, smartwatch-based detection could become a valuable tool in emergency cardiac care, offering life-saving intervention when seconds matter.
Katrina Thornber, EMJ
Reference
Shah K et al, Automated loss of pulse detection on a consumer smartwatch. Nature. 2025;DOI:10.1038/s41586-025-08810-9.
