Fig. 5. Effects of ventilator mode and level of assistance on the coefficient of variation (CV) and autocorrelation analysis for representative breathing pattern variables. Compared with PSV, values for the CV are somewhat greater during NAVA and rise further as the gain increases from 1 to 4 cm H2O/μV diaphragm electrical activity. Autocorrelation analysis is used here to determine the number of breaths through which a significant correlation or lag is found. Considered an index of “short term memory,” the greater the numerical value of lag, the larger the duration of interbreath correlation. With the exception of mean inspiratory flow, lag seems to be lower during NAVA. Overall, these results are consistent with increased breathing pattern variability during NAVA when compared with PSV. CV = coefficient of variation; CVRR= CV of respiratory rate; CVVt= CV of tidal volume; CVTi/Tt= CV of the inspiratory duty cycle; CVVt/Ti= CV of mean inspiratory flow; lag = number of breaths among which a significant correlation was found; lagVt= lag of tidal volume; lagRR= lag of respiratory rate; lagVt= lag of the tidal volume; lagTi/Tt= lag of the inspiratory duty cycle; lagVt/Ti= lag of mean inspiratory; NAVA1,2,3,4= neurally adjusted ventilatory assist with a gain of 1, 2, 3, 4 cm H2O/μV diaphragm electrical activity flow; PSV = pressure support ventilation. Data are means ± SEM; *P < 0.05.