Online estimation of respiratory mechanics in non-invasive pressure support ventilation: a bench model study.

An online algorithm for determining respiratory mechanics in patients using non-invasive ventilation (NIV) in pressure support mode was developed and embedded in a ventilator system. Based on multiple linear regression (MLR) of respiratory data, the algorithm was tested on a patient bench model under conditions with and without leak and simulating a variety of mechanics. Bland-Altman analysis indicates reliable measures of compliance across the clinical range of interest (± 11-18% limits of agreement). Resistance measures showed large quantitative errors (30-50%), however, it was still possible to qualitatively distinguish between normal and obstructive resistances. This outcome provides clinically significant information for ventilator titration and patient management.

Automated detection of asynchrony in patient-ventilator interaction.

An automated classification algorithm for the detection of expiratory ineffective efforts in patient-ventilator interaction is developed and validated. Using this algorithm, 5624 breaths from 23 patients in a pulmonary ward were examined. The participants (N = 23) underwent both conventional and non-invasive ventilation. Tracings of patient flow, pressure at the airway, and transdiaphragmatic pressure were manually labeled by an expert. Overall accuracy of 94.5% was achieved with sensitivity 58.7% and specificity 98.7%. The results demonstrate the viability of using pattern classification techniques to automatically detect the presence of asynchrony between a patient and their ventilator.

Automatic detection of ineffective triggering and double triggering during mechanical ventilation.

OBJECTIVE: Imperfect patient-ventilator interaction is common during assisted ventilation, and the detection of clinically relevant mismatching requires visual monitoring of the ventilator screen. We have assessed the feasibility, sensitivity and specificity of an algorithm embedded in a ventilator system that is able to automatically detect the occurrence of ineffective triggering and double triggering in real time. DESIGN: Prospective study. SETTING: Respiratory intensive care unit. METHODS: Twenty patients undergoing pressure-support ventilation, either non-invasively (NIV, n=10) or conventionally ventilated (n=10), were studied. MEASUREMENTS: The detection of ineffective triggering and double triggering from the algorithm was compared by two operators with the "real" occurrence of the phenomena as assessed using the transdiaphragmatic pressure (Pdi). RESULTS: Seven of the 20 patients exhibited gross mismatching, while in the remaining patients patient-ventilator mismatching was artificially induced using a pressure control, with a low respiratory rate. Ineffective triggering and double triggering were identified by the operators in 507 and 19 of the 3343 analyzed breaths, respectively. False positives were significantly more frequent in the NIV group than with conventional ventilation. The algorithm had an overall sensitivity of 91% and specificity of 97%. Specificity was statistically higher in the conventional ventilated group than with NIV (99% vs. 95%, p<0.05). CONCLUSIONS: We have demonstrated the feasibility and efficacy of a new algorithm to detect the occurrence of impaired patient-ventilator interaction during mechanical ventilation in real time. This software may help the clinician in the identification of this problem, which has been shown to have important clinical consequences.