Harmonic analysis of pulse morphology variability for pulse smoothness assessment

A non-linear physiological system leads to a change in pulse morphology variability (PMV), which can be an important feature in assessing the condition of the cardiovascular system. The smooth pulse derived from traditional East Asian medicine (TEAM) also contains chaotic features according to the temporal sequence, and such pulses usually appear during menstruation in women. Analyzing the PMV as a novel indicator of pulse smoothness can contribute not only to understanding non-linear physiological systems but also to studying the characteristics of pulse patterns. In this study, we propose an algorithm to assess pulse smoothness using a harmonic analysis approach of the PMV. First, we introduced a two-step pre-processing method that considers the applied pressure (AP) variability and outlier pulse removal (OPR) to generate a refined pulse series. Next, we performed PMV analyses using four different methods to examine the characteristics of the pulse series. Finally, we performed a spectral harmonic analysis based on the trace of the intra-class distance within each single-period pulse (TIS) to assess the pulse smoothness. We evaluated the proposed algorithms using repeated-measures ANOVA and receiver operating characteristic (ROC) analysis according to the menstrual period. Distorted pulses were automatically detected with the pre-processing method, and the maximum amplitude of the average TIS was consistently observed near the radial augmentation index point. In addition, the total adjacent harmonic peak increment (AHPI) among the proposed variables was significantly higher during the menstrual phase than during the non-menstrual phase (P < 0.05), and the area under the ROC curve of the AHPI was 0.742. Therefore, dissonant or disharmonic frequency components are present during menstruation, and the AHPI could be a novel indicator reflecting pulse smoothness.