Sensitivity analysis of a computer model of neonatal oxygen transport

Abstract

Computer models of neonatal oxygenation could serve as a tool for a comprehensive comparison of closed-loop automated oxygen control systems. The behaviour of such models depends, besides the input data of the inspired fraction of oxygen and the premature infant's breath pattern, on internal parameters of the model. The aim of this study was to perform a sensitivity analysis of a computer model of neonatal oxygen transport to clarify the influence of its internal physiological parameters on the output signal of peripheral oxygen saturation (SpO2). We performed a multi-parameter sensitivity analysis using Monte Carlo simulations for randomly generated values of eight internal parameters. The influence on the model output SpO2 signal was evaluated using five characteristics of the output signal. The relations between the parameters and the output characteristics were displayed using scatter plots and analysed by linear correlation, standardized regression, and partial correlation. The main result of the study is that in our model the oxygen consumption in the tissue and the cardiac output have the greatest influence on the SpO2 drop and minimal SpO2 value during simulated desaturation. The rate of development of desaturation and its duration are most affected by the diffusion resistance of the alveolar-capillary membrane. The results of the sensitivity analysis will help to optimize the performance of the computer model of neonatal oxygen transport.

Keywords

• automated control
• computer model
• peripheral oxygen saturation
• preterm infant
• sensitivity analysis