Abstract
Analysis of electrogastrographic (EGG) signals requires methods capable of characterizing structural organization across scales while retaining temporal-order information. We present a multiscale approach based on horizontal visibility graphs (HVGs), consecutive outerplanar motifs, grouped probability distributions (GPDs), and information-theoretic quantifiers. The methodology is illustrated using two representative 28-sample EGG segments and subsequently evaluated using 19 complete recordings across motif sizes m = 3,..., 50. Finite-sample effects were assessed by progressively varying signal length, the grouped representation was compared with ungrouped HVG topology counts, and recording-level grouped-HVG quantifiers were compared with conventional EGG spectral descriptors. Finite-sample deviations progressively decreased with increasing signal length, whereas the ungrouped representation exhibited rapid growth in the number of observed topological states at larger motif sizes. GHVGSC exhibited exploratory negative correlations with the percentage of normogastria at m = 42 − 50, which were significant under within-family BH-FDR correction but did not survive global correction across all 384 tests. Overall, the proposed framework provides a reproducible multiscale characterization of EGG structural organization, while the expanded evaluation establishes its finite-sample behavior and its relationships with ungrouped HVG representations and conventional spectral descriptors.