A Priori Identifiability Analysis of Cardiovascular Models.

Model parameters, estimated from experimentally measured data, can provide insight into biological processes that are not experimentally measurable. Whether this optimized parameter set is a physiologically relevant complement to the experimentally measured data, however, depends on the optimized parameter set being unique, a model property known as a priori global identifiability. However, a priori identifiability analysis is not common practice in the biological world, due to the lack of easy-to-use tools. Here we present a program, Differential Algebra for Identifiability of Systems (DAISY), that facilitates identifiability analysis. We applied DAISY to several cardiovascular models: systemic arterial circulation (Windkessel, T-Tube) and cardiac muscle contraction (complex stiffness, crossbridge cycling-based). All models were globally identifiable except the T-Tube model. In this instance, DAISY was able to provide insight into making the model identifiable. We applied numerical parameter optimization techniques to estimate unknown parameters in a model DAISY found globally identifiable. While all the parameters could be accurately estimated, a sensitivity analysis was first necessary to identify the required experimental data. Global identifiability is a prerequisite for numerical parameter optimization, and in a variety of cardiovascular models, DAISY provided a reliable, fast, and simple platform to provide this identifiability analysis.