Rule-based Arrhythmia Screening from Single-Lead ECG for Continuous Monitoring

Cardiac arrhythmias often occur intermittently and may be missed during short clinical examinations, motivating the need for continuous monitoring solutions. Single-lead ECG recordings provide a practical approach for prolonged monitoring, but are affected by noise, motion artifacts, and inter-subject variability, which can impact the reliability of existing methods. This paper presents a rule-based framework for arrhythmia screening from single-lead ECG signals. The method combines temporal features derived from RR interval dynamics with morphological descriptors of the QRS complex within a transparent window-based scoring scheme, and introduces an aggregation strategy across partially overlapping windows to support reliable screening of arrhythmic patterns at the recording level. The approach is evaluated on 89 ECG recordings from both clinical and real-world datasets, achieving 89.9% accuracy, 93.8% sensitivity, and 85.4% specificity. These findings support the use of single-lead ECG for arrhythmia screening under heterogeneous acquisition conditions and suggest the potential applicability of the proposed framework to wearable and textile-based ECG systems, where non-invasive acquisition and variable signal quality must be effectively handled.