Parameter Identification Methods in a Model of the Cardiovascular System

To be clinically relevant, mathematical models have to be patient-specific, meaning that their parameters have to be identified from patient data. To achieve real time monitoring, it is important to select the best parameter identification method, in terms of speed, efficiency and reliability. This work presents a comparison of seven parameter identification methods applied to a lumped-parameter cardiovascular system model. The seven methods are tested using in silico and experimental reference data. To do so, precise formulae for initial parameter values first had to be developed. The test results indicate that the trust-region reflective method seems to be the best method for the present model. This method (and the proportional method) are able to perform parameter identification in two to three minutes, and will thus benefit cardiac and vascular monitoring applications.