Analysis of the Method for Determining Changes in the Airways from the Spirometric Curve Evolution

Abstract

The severity of chronic respiratory diseases is evaluated performing spirometry, and particularly the forced expiration maneuvers before and after bronchodilation or challenge tests. However, no method has yet been proposed for the quantitative assessment of changes in airway mechanics following such tests. Just recently, a reduced model for forced expiration with 6 free parameters was derived and used to estimate the parameters by fitting it to a spirometric curve. The aim of this work was to perform comprehensive research on the method for quantifying the changes in airway mechanics by fitting the above model to two spirometric curves, representing the states of the respiratory system before and after bronchodilation or bronchoconstriction. To this end, a set of pairs of spirometric curves were generated from randomly drawn parameters, and 2,400 of them were used for testing purposes. The proposed method for spirometric data analysis consisted of two stages: the estimation of 6 parameters from the pre-test data using the inverse neural network and the Levenberg-Marquardt (LM) algorithm, and then the estimation of 2 parameters describing airway properties from the post-test curve with the LM procedure. The results show that this approach allows the quantification of changes in the airway mechanics with the accuracy of about 6-7 % of the parameter ranges. These outcomes encourage further analysis of the method using more reliable spirometric data.