Hybrid Amplitude Ordinal Partition Networks for ECG Morphology Discrimination: An Application to PVC Recognition.

Various algorithms have emerged for automatic electrocardiogram analysis, focusing on discerning subtle changes in arrhythmic morphology, especially premature ventricular contractions. While ordinal partition networks have been effective in analyzing real-world time series data, conventional ordinal partition networks primarily concentrate on local electrocardiogram shapes, overlooking amplitude-level information. This study introduces an innovative method, absolute amplitude ordinal partition networks, which incorporates amplitude-level variations (coarse-grained electrocardiogram) into ordinal patterns. These absolute amplitude ordinal partition networks encompass all conceivable ordinal patterns as nodes, connected based on temporal sequences. Ten network measures are extracted from both ordinal partition networks and absolute amplitude ordinal partition networks, capturing shape and level-based premature ventricular contractions characteristics. These measures are integrated into hybrid amplitude ordinal partition networks to construct Support Vector Machine-based premature ventricular contractions recognition models. Evaluated on three baseline databases (Massachusetts Institute of Technology-Beth Israel Hospital arrhythmia database (96,587 beats), St. Petersburg Institute of Cardiological Technics database (156,373 beats), and China Physiological Signal Challenge 2020 database (987,209 beats)), the proposed models demonstrate F1-scores of 97.02%, 93.06%, and 91.03% across databases in class-oriented evaluation, and 94.57%, 87.96%, and 88.89% in subject-oriented evaluation. Notably, the model achieves premature ventricular contractions scores of 53,904 and 59,201 on the China Physiological Signal Challenge 2020 test set for the two evaluations, affirming its efficacy in premature ventricular contractions recognition. This framework provides a versatile approach for detecting morphology anomalies in various physiological signals, such as heart sounds and pulses.