Fig 1.
Pulse wave analysis technique.
Non-invasive pulse wave recording in peripheral artery is used to synthesize the aortic pressure wave using a generalized transfer function. Characteristic points in the aortic pressure waveform are then used to calculate various indices related to the cardiovascular system state.
Table 1.
Basic hemodialysis (HD) and control groups characteristics before and after age and gender based propensity score matching (PSM).
Reported values for HD group were assessed after midweek hemodialysis session. The data about 20 out of 32 control subjects have been published previously in [20].
Fig 2.
(A) Eight separate recordings of radial pulse wave were performed for each dialysis patient. In healthy individuals recording was performed only once. (B) Each radial pressure recording was used to estimate parameters of the pulse wave propagation model and derive patient-specific information about cardiovascular system state.
Fig 3.
Comparison of the model predicted and measured pressure profiles in radial artery.
Bars represent average model-predicted pressures at different measurement moments after 3- (A) and 2-day (B) interdialytic break. Shown are also inter-patient means and standard deviations for measurements (circles with whiskers) whereas standard deviation from the model is omitted for clarity. Asterisks indicate statistical difference at p-value < 0.05 calculated using paired sample t-Test. (C) Quality of the model approximation, i.e. average relative error between simulated and recorded radial pressure waveform, does not depend on the measurement moment.
Fig 4.
Effects of hemodialysis on the model-estimated parameters.
Repeated measurements statistical testing revealed that hemodialysis affects the model-predicted stiffness of large arteries (A), stroke volume (B), and peak heart ejection moment (C). Each box represents the interquartile range with the circle within being the sample median. Hollow circles denote outliers, i.e. values that are more than 1.5 times the interquartile range away from the top or bottom of the box. Whiskers show the furthest observations when neglecting outliers. The global p-value from the Wittkowski test statistics is shown above each boxplot. Solid red line with an asterisk above indicates multiple-comparison p-value < 0.05 whereas dotted line shows borderline statistical significance, i.e. p-value < 0.1.
Fig 5.
Comparison of the control and hemodialysis (HD) groups after propensity score matching.
First column shows average values of the model-estimated (names in red font) and SphygmoCor-derived (names in black font) parameters in control group. Colormap encodes the percent change in the average parameter value for hemodialysis patients compared to healthy counterparts. Exact percent change value is shown in the cases for which the difference is statistically significant (diamond for p-value < 0.05 and square for borderline p-value < 0.1).
Fig 6.
Interdependence of the clinical and pulse wave-derived characteristics in hemodialysis patients.
Color coded are the Spearman correlation coefficients for four different pulse wave measurement moments. Font color encodes the type of the parameter, i.e. whether it is model-derived (red), SphygoCor-derived (black), or it is a basic clinical characteristic (blue). White squares indicate lack of statistical significance. Three black squares highlight the main correlation clusters.