Accurate Identification of Actionable Single-Lead ECG Data Using a Signal Quality Assessment Algorithm

Accurate identification of high-quality ECG data is crucial for reliable cardiac rhythm assessment, particularly in ambulatory settings.This study presents a machine learning-based signal quality assessment algorithm designed to address the challenges of ECG signal classification in diverse and real-world environments.The algorithm prioritises sensitivity, effectively identifying actionable ECG segments while minimising the risk of discarding critical clinical data.Trained and validated on a wide range of public and proprietary ECG datasets, the algorithm maintains a consistently high sensitivity (>98%) across various device types, patient demographics, and recording environments.Our results highlight the algorithm's utility as a generalisable signal quality triage tool, with the potential to reduce burden on clinicians and enhance the efficiency of ECG analysis in clinical practice.