Cardiac Function Assessment Using Personalized Reduced-Order Pulse Wave Modeling

Kamil Wołos and Jan Poleszczuk
Nalecz Institute of Biocybernetics and Biomedical Sciences, Polish Academy of Sciences


Background: Currently, invasive and imaging methods are the primary diagnostic tools for cardiovascular diseases. Personalized 0-1D dimensional models of the cardiovascular system may be used as an auxiliary tool for cardiac function assessment.

Objective: We set out to investigate the possibility of estimating heart-related indices such as stroke volume or shape of the left ventricle elastance function using non-invasive measurement of the pulse wave from the radial artery.

Methods: Peripheral pressure waveforms were recorded using applanation tonometry (SphygmoCor, AtCor, Australia) in 35 hemodialysis patients and used to estimate subject-specific parameters of a 1D pulse wave propagation model coupled with the 0D elastance-based model of the heart. Bioimpedance cardiography measurements (PhysioFlow, Manatec Biomedical, France) were also performed to cross-validate model predicted cardiac function parameters.

Results: Pulse waveforms have been well reproduced by the model, with a mean absolute percentage error 4.09%. We found interesting significant correlations between parameters describing the shape of the elastance function and aortic indices, achieving R values above 0.5 with a p-value < 0.01. We also noted that the best correlation between computed and bioimpedance-measured stroke volume occur for recordings performed after a shorter interdialytic break, before the start of the hemodialysis session (R = 0.72 with p-value < 0.001).

Conclusion: The proposed model can reflect well the recorded pulse waveforms and can provide information about some of the cardiac function parameters.