Evaluating cardiological rehabilitation using wavelet analysis: a pre- and post-rehabilitation comparison

Abstract

Abstract Cardiac rehabilitation (CR) improves prognosis and exercise capacity, but conventional indices may not fully capture subtle changes in cardiovascular signal dynamics. Time–frequency analysis of cardiovascular signals may provide additional, candidate time–frequency features of cardiovascular signal dynamics. We studied 20 patients (17 men, 62 ± 9 years) undergoing a standard ~5-week outpatient cardiological rehabilitation program. Symptom-limited exercise tests were performed before and after CR to assess maximal power, exercise duration, resting and maximal heart rate, and heart rate recovery time. At rest, continuous ECG and non-invasive arterial blood pressure (BP) were recorded in the supine position. Wavelet transform (WT) was applied to derive time-averaged amplitudes in five predefined frequency intervals (I–V) and to estimate wavelet phase coherence and phase difference between ECG and BP. Pre–post differences were evaluated using paired Wilcoxon signed-rank tests, whereas intersubject surrogate distributions were used separately to assess whether within-subject ECG–BP coherence exceeded an empirical null threshold. Between the pre- and post-rehabilitation assessments, maximal power increased significantly (median 103 → 129.5 W; p = 0.000576), whereas no statistically significant pre–post differences were detected in exercise duration, resting or maximal heart rate, or heart rate recovery. In the normalized CNAP-derived peripheral pressure signal, interval IV showed a statistically significant pre–post increase after Holm correction, whereas the interval V difference did not remain significant after adjustment. In the complementary non-normalized analysis, no significant increase was detected in intervals IV or V; instead, interval II showed a significant decrease after Holm correction. Thus, the normalized interval-IV finding represents a change in relative frequency-specific signal content rather than evidence of an absolute increase in BP oscillatory amplitude. Previous microvascular and cardiovascular studies have associated intervals IV and V predominantly with neurogenic vasomotor modulation and nitric oxide-related endothelial activity, respectively; however, the underlying physiological mechanisms were not directly assessed in the present study. No statistically significant pre–post differences were detected in the analyzed ECG–BP phase coherence intervals (I and II) or in phase difference across intervals I–V. Wavelet analysis identified a statistically significant pre–post difference in the normalized CNAP-derived peripheral pressure signal within interval IV after correction for multiple comparisons. A nominal difference was also observed in interval V, but it did not remain statistically significant after Holm correction. Because these mechanisms were not measured directly in the present cohort, these associations provide a physiological framework for interpretation of the observed oscillations rather than direct evidence of changes in autonomic function. These exploratory findings characterize pre–post differences in cardiovascular signal dynamics; their reproducibility, physiological significance, and incremental clinical value relative to conventional assessments remain to be established.

The authors' abstract, as published at the source. The European Physical Journal Special Topics, 2026 · DOI ↗