Optimising implantable loop recorder alerts with artificial intelligence: role of R-wave amplitude and workload reduction.

Abstract

Background: False-positive (FP) alerts from implantable loop recorders increase clinical workload and may delay appropriate intervention. AccuRhythm AI, a cloud-based filtering algorithm, is designed to reduce these alerts in Reveal LINQ and LINQ II devices. This study assessed the algorithm's effect on FP and clinician burden reduction, with a focus on the influence of R-wave sensing amplitude.

Methods: This multicentre, retrospective study included 800 patients with either Reveal LINQ or LINQ II. We analysed automated artificial intelligence (AI) reports and compared FP rates and transmission burden before and after software-based AI activation in the subset of Reveal LINQ patients to assess patient-level changes. The relationship between R-wave amplitude and FP incidence was also evaluated.

Results: AI-based filtering, by AccuRhythm AI automatic analysis, reduced false pause alerts by 62% and false atrial fibrillation alerts by 33%, saving 210 clinician hours over 6 months. Patient level analysis, among 465 Reveal LINQ patients, showed FP+ patients (patients with ≥1 false-positive transmission) reduction from 55.5% to 15.1% post-AI (p<0.001), translating to 1128 hours saved. All residual false alerts occurred in patients with R-wave amplitudes <0.4 mV.

Conclusion: Use of AccuRhythm AI was associated with a significant reduction in FPs and clinician workload while preserving diagnostic accuracy. R-wave amplitude remained a key factor influencing alert specificity, emphasising the continued importance of optimal device implantation and signal quality.