Negative impact of the noseclip on high-frequency respiratory impedance measurements

A recurrent parameter model to characterize the high-frequency range of respiratory impedance in healthy subjects

In this work, a re-visited model of the respiratory system is proposed. Identification of a recurrent electrical ladder network model of the lungs, which incorporates their specific morphology and anatomical structure, is performed on 31 healthy subjects. The data for identification has been gathered using the forced oscillation lung function test, which delivers a non-parametric model of the impedance. On the measured frequency response, the ladder network parameters have been identified and a fractional order has been calculated from the recurrent ratios of the respiratory mechanics (resistance and compliance). The paper includes also a comparison of our recurrent parameter model with another parametric model for high frequency range. The results suggest that the two models can equally well characterize the respiratory impedance over a long range of frequencies. Additionally, we have shown that the fractional order resulting from the recurrent properties of resistance and compliance in the ladder network model is independent of frequency and is not biased by the nose clip wore by the patients during measurements. An illustrative example shows that our re-visited model is sensitive to changes in respiratory mechanics and the fractional order value is a reliable parameter to capture these changes.

Computer assessment of indirect insight during an airflow interrupter maneuver of breathing

The paper answers the questions if it is possible to conclude in objective way on more (than one -Rint - in a classical IT) number of parameters from the time domain post-interrupter signals during the occlusional measurement of respiratory mechanics and also verifies what accuracy can be achieved in such attempt. To obtain reported results, the time-domain enhanced interrupter technique (TD-EIT) was developed in this paper using computer simulations. Three-stage scheme of work was assumed in the project. First, the quality of the model identification was assessed for various combinations of pressure and flow signals recorded during the interruption. Then, the correlation between the working characteristics of the interrupter valve and the precision of the parameter estimation were assessed for the TD-EIT algorithm. Finally, a verification experiment by forward-inverse modeling was organized, in which the mechanical characteristics of a complex model were mapped with reduced analogs and with the use of neural networks for three typical modes: 'Normal state', 'Airway constriction' and 'Cheeks supported'. Obtained results show that to became effective in time-domain post-interrupter data exploration, both pressure and flow signals should be used in assessment of respiratory mechanics, taken in a range of at least 100ms and when both slopes (valve closing and opening) of quasi-step excitation are included. What is more, the faster the valve the smaller error of parameter estimation in proposed TD-EIT was observed, and this uncertainty importantly falls down for the length of time window exceeding the limit of 100ms. The pioneering use of neural network for mapping the mechanical properties of lungs with the use of interrupter experiment methodology proves that it is possible to conclude about more (than one) number of parameters characterizing the complex system and that this insight is biased with the error not exceeding of 10%; only peripheral properties are estimated worse. Such observation has a potential to change the experimental protocol, which was used in interrupter measurements up to date and to make this technique more attractive in comparison to other method, i.e. forced oscillation technique or impulse oscillometry. As regards the practical meaning of reported results for engineers and end-users (physicians and patients), proposed solution can be applied in simple portable devices with a feature of easy operation (important for e-monitoring).