Amsterdam-based Philips presented an abstract highlighting recent research at the annual Hearth Rhythm Society (HRS) congress in New Orleans. The study, “Near-Term Prediction of Life-Threatening Ventricular Arrhythmias using Artificial Intelligence-Enabled Single Lead Ambulatory ECG,” used an AI-based learning model. It successfully predicted sustained ventricular tachycardia (VT) over a two-week period when compared to ambulatory ECG data.
Philips developed its AI-based learning model using a deep neural network and 115,505 ambulatory ECG recordings. The company collected these recordings from independent diagnostic testing facilities across five countries. It then retrospectively validated the model using retrospective 14-day ambulatory ECG recordings from 2019 to 2023.
Researchers studied the algorithm’s ability to predict the risk of sustained VT (lasting longer than 30 seconds). The study evaluated a two-week period using data from the first 24 hours of monitoring.
The model achieved a sensitivity of 83.3% and a specificity of 88.7% on its internal validation dataset. On the external validation dataset, sensitivity and specificity came in at 78.9% and 81.4%, respectively. The model correctly predicted VT occurrence in 88% of Holter users with rapid VT.
Philips said novel AI models “lay the foundation for a new approach to cardiac risk management.” The company believes it can apply across multiple care settings, helping to identify at-risk patients earlier.
“AI-powered digital biomarkers have the potential to advance cardiac care pathways by moving from reactive to preventive medicine,” said Manish Wadhwa, head of medical office, ambulatory monitoring and diagnostics at Philips. “Predictive biomarkers may enable early risk detection, enhanced patient monitoring, and improved patient management, helping to facilitate better outcomes.”