Inverse estimation of model parameters for newborn brain cooling process simulations
In this work, a three-dimensional simplified computational model was built to simulate the passive thermophysiological response of part of a newborn's head for neonate’s selective brain cooling. Both metabolic heat generation and blood perfusion were considered. The set of model parameters was selected and a sensitivity study was carried out. Analysis of dimensionless sensitivity coefficients showed that the most important are: the contact thermal resistance between the cool-cap and skin, the thermal resistance of the plastic wall material, and deep (arterial) blood temperature.
The function specification method was applied to estimate the value of the contact resistance. Two, four and six computationally simulated measurements with different uncertainties were used to adjust random contact resistance value to the assumed one. Results showed that when using only two measurements having 2% of the uncertainty, the error of estimation does not exceed 9.8%. However, when using six measurements having 1% of uncertainty, the resulting estimation is burdened with an error of 0.3% only.