Abstract Or103: Automated cardiac arrest detection incorporated into a wristband: validation in patients with induced ventricular fibrillation

Background: Survival from unwitnessed out-of-hospital cardiac arrest is poor, often due to delayed activation of the emergency medical chain. Wearable technology capable of automated cardiac arrest detection and alerting could trigger rapid medical assistance. In a previous study, we developed an algorithm for cardiac arrest detection based on photoplethysmography (PPG) data from patients with induced circulatory arrest, achieving 98% sensitivity. The next step is validation of the algorithm in patients with shockable cardiac arrest. Goal: To study the performance of the developed PPG-algorithm in patients with cardiac arrest based on induced ventricular fibrillation (VF). Methods: From the prospective multicenter study DETECT-1, we selected all adult patients who underwent short-lasting VF induction as standard practice during subcutaneous implantable cardioverter defibrillator (S-ICD). Patients wore a PPG-wristband (CardioWatch) during the entire procedure. A cardiac arrest event was defined as induced VF. Continuous electrocardiogram (ECG) and arterial blood pressure were monitored as a reference standard. The primary endpoint was the sensitivity for the detection of cardiac arrest, as assessed using the previously developed PPG-algorithm. This PPG-based algorithm detects the absence of pulsations based on signal amplitude and waveform characteristics. Results: In total, 14 patients were included with a median age of 46 years (IQR 33-54), of whom four where female. Fifteen VF inductions were performed, see Figure 1. The sensitivity for cardiac arrest detection was 100% (95% confidence interval [CI] 75%-100%). No false positive cardiac arrest alerts occurred in 15 hours of PPG data, resulting in a positive predictive value of 100% (95% CI 75%-100%). Return of PPG pulsations after the S-ICD shock were detected successfully in all patients. Conclusions: The PPG-algorithm for cardiac arrest detection performs excellent in the detection of induced VF. As a next step, the algorithm needs to be validated in non-shockable cardiac arrest and potential false positive alarms during daily life use need to be studied.