Algorithm Improvements in an Insertable Cardiac Monitor Reduce False Positives and Episode Review Burden

Introduction The use of insertable cardiac monitors (ICM) for arrhythmia monitoring continues to grow steadily. However, ICM performance remains less than perfect, and the volume of transmitted data poses a challenge to clinics. Here, we evaluate the impact of three improvements implemented in the LUX‐Dx II+ ICM for atrial fibrillation (AF), pause, and bradycardia detection. Methods Enhanced AF and pause algorithms were evaluated using real‐world ECG data and compared against the predecessor LUX‐Dx algorithms. Optimized nighttime (11 p.m.–7 a.m.) programming settings were simulated on LUX‐Dx data to evaluate the reduction of non‐actionable nocturnal bradycardia (>30 bpm) and pause (<5 s) detections. Results The novel AF algorithm reduced false positives by 34% in the training dataset and 38% in the test dataset while maintaining >98% sensitivity relative to the legacy algorithm. False positive reduction and relative sensitivity were even higher (74% and 100%, respectively) in a subset of patients with more sensitive AF programming settings. The enhanced pause algorithm reduced false positives by 62.5% in the training set and 48.6% in the test set, while maintaining 100% relative sensitivity. Application of adjusted nocturnal detection thresholds reduced the number of nighttime bradycardia and pause episodes by 98% and 90%, corresponding to an overall episode reduction of 75% and 56%, respectively. Conclusion The enhanced ICM algorithms and programming settings substantially reduced false positive AF and pause detections as well as non‐actionable nighttime bradycardia and pause detections. These enhancements are expected to improve ICM performance and workflow in the device clinic.