Identifiability of left ventricular end-systolic pressure-volume relationships.

Abstract

A widely accepted model of the left ventricle (LV) consisting of a time-varying elastance and a nonlinear internal resistance was investigated to make inferences about the identifiability of its parameters by means of simulated experiments. We aimed to retrieve maximum elastance (Emax) and dead volume (Vd) by the usual slope method or end-systolic pressure-volume relations (ESPVR) and by model-based parameter identification. The ESPVR deviated increasingly from the assigned values with increasing internal resistance depending on the type of loading intervention. Model-based parameter identification proved to be hampered by considerable error propagation if applied to single contractions with noise on the data. Better results were obtained by reducing the number of parameters to be estimated or by combining contractions with different loading conditions. The LV model was also matched with experimental data in three open-chest anesthetized dogs when both methods of estimation were used. The trend of ESPVR was in accordance with the model predictions, with larger Emax and larger Vd observed with arterial rather than with venous loading. Inclusion of an internal resistance in the classical elastance model can explain the dependence of the ESPVR on the type of loading intervention. However, application of model-based parameter identification indicates that the model fails to represent the entire systolic pressure-volume time course of the in situ LV.