Population representativeness and optimality of fitted PDFs for coronary artery SEF parameters

Determine whether the probability density functions fitted to layer-specific strain energy function parameters (u, k1, k2, fiber angle φ, and fiber dispersion p) of human coronary arteries—using gamma distributions for u, k1, k2, and φ, and a beta distribution for p based on mechanical testing of 13 nonatherosclerotic left anterior descending coronary artery samples—accurately represent population-level parameter distributions and whether these chosen distribution families are the optimal descriptors of the true parameter variability.

Background

The paper constructs an uncertainty quantification pipeline by fitting univariate probability distributions to strain energy function parameters for the media and adventitia layers of human coronary arteries, derived from layer-specific mechanical tests on 13 samples. Gamma distributions were used for u, k1, k2, and φ, while a beta distribution was used for p, under an independence assumption.

While these fitted PDFs capture the observed sample distributions, the limited sample size and the selection of distribution families raise questions about whether they adequately reflect population-level variability. The authors explicitly note uncertainty regarding whether these fits represent the true population distributions and whether the selected distributions are optimal.