On-line identification of mechanical parameters of the lung

Abstract

The subject of this paper is the use of a microcomputer for on-line estimation of mechanical parameters of the lungs. The identification is based on measurements of airflow at the mouth using a pneumotachograph and oesophageal pressure obtained via a balloon catheter. A simple mathematical model for lung mechanics is used which includes terms for airways resistance, dynamic pulmonary compliance and a non- linear turbulence factor. The method of identification is a steepest-descent gradient algorithm based on equation- error formulation of the problem. The stability of the method has been verified via Liapunov theory, and the tracker was originally implemented in analogue form. It has been shown to give good dynamic tracking performance, with settling times in the order of a few breathing cycles. Good correlations between manual off-line estimation and the on-line steepest-descent tracker have been obtained in both human and guinea-pig trials. Recently, the estimation algorithm has been transferred to a Research Machines 380Z microcomputer and applied to neo-natal data for rapid assessment of apnoea. Two versions of programmes have been implemented, one of which gives an on-line printer listing of lung parameters based on either inspiratory or expiratory data. The other version uses four DA C channels to give analogue chart recording data Examples are shown of simultaneous six-channel recordings comprising the primary pressure and flow signals, estimated airways resistance and pulmonary compliance, tidal volume and minute volume. The effect of apnoea on these parameters, together with heart rate, can be assessed using this on-line identification method.