Evaluation of a mock circulation model through time-series analysis.

Abstract

A mathematical model of the Donovan Mock Circulation Unit (MCU), derived from fundamental principles of fluid dynamics is utilized to represent this system. A new solution to a lumped-parameter model of the human circulation is developed using discontinuous line segments to approximate the complicated aortic flow profile. Aortic pressure and flow data for the human circulation are taken from the literature and are collected as a time-series from the mock circulation unit. The model is fitted to this data using non-linear regression analysis. Residual plots and autocorrelation functions for the data sets suggest that the model should be fitted using a first-order moving average process and tests for lack-of-fit indicate that the model is adequate. Parameter estimates are within acceptable physiological limits. Approximate inference bands for the model predictions are presented for both the human physiological data and the MCU data. (Abstract shortened by UMI.)