Pause-triggered episode classification algorithms and high-priority review queue concept for cardiac monitoring devices

Abstract

Abstract Background Cardiac monitoring technologies often utilize monitoring centers or services in which technicians adjudicate arrhythmia episodes. Arrhythmias with a high prevalence of false detections (such as pauses) can slow down the review process while also delaying review of time-sensitive true episodes. Objective To develop algorithms that can stratify device-detected pause episodes into low- and high-priority queues to facilitate monitoring center review. Methods The high priority queue algorithm identifies episodes which are likely to be true pause episodes. The algorithm consists of four conditions, which identify features that were found to be predictive of true pauses, including the number of beats in the episode, the noise status of the last pre-pause beat, and the relative flatness of the ECG signal. Using a similar set of features, a set of criteria was determined that would identify pause-triggered episodes that were highly unlikely to be true pauses. 11,567 pause episodes were used for development with 19,520 separate episodes used for validation. All episodes were adjudicated by a cardiac monitoring center. The validation dataset consisted of 18,280 (93.6%) false pauses and 1240 (6.4%) true pauses. Results The high-priority queue algorithm identified true pause episodes with a sensitivity of 82.3% and specificity of 96.8% in the validation dataset (see table). The low-priority queue algorithm flagged 78.4% of all pause episodes as low-priority in the validation dataset, with only 4 true pause episodes (<0.1%) flagged as low priority. Conclusion The high-priority queue algorithm for device detected pause episodes could potentially identify ∼82% of all true pause notifiable episodes, expediting their review process. The low-priority queue successfully identifies a large percentage (∼80%) of false pause-triggered episodes, which would help to improve monitoring center efficiency as false pause-triggered episodes are a large driver of artifact / normal sinus rhythm episodes. Funding Acknowledgement Type of funding source: Private company. Main funding source(s): Medtronic