Monitoring variability in parameter estimates for lumped parameter models of the systemic circulation using...

Area of Science:

• Cardiovascular Physiology
• Computational Modeling
• Biomedical Engineering

Background:

• Physics-based cardiovascular models are emerging for clinical disease diagnosis and prognosis.
• Model personalization requires accurate physical and physiological parameters.
• Individualized parameters offer insights into specific patient states and disease etiology.

Purpose of the Study:

• To apply a fast model optimization scheme to personalize cardiovascular models.
• To investigate two model formulations (closed-loop and open-loop) of the left ventricle and systemic circulation.
• To assess the feasibility of distinguishing individual parameter estimates from population estimates.

Main Methods:

• Utilized a local optimization scheme for model personalization.
• Applied both closed-loop and open-loop cardiovascular models.
• Used intermittently collected hemodynamic data (brachial pressure, stroke volume, outflow tract velocity) from 25 participants across multiple measurement days.
• Paired pressure waveforms from either the finger artery or carotid artery.

Main Results:

• Estimated parameter values were similar between the two model formulations.
• Systemic arterial compliance estimates significantly differed based on the choice of pressure waveform (finger vs. carotid).
• Higher systemic arterial compliance estimates were observed using the finger artery pressure waveform.

Conclusions:

• Parameter estimate variability within a participant on a single day was lower than across multiple days or the population.
• The optimization method successfully distinguishes individual participants from the general population.
• Different measurement days for individual participants can be identified by their parameter values.