Neonatology Department, University of Liège, Belgium.
Pediatr Crit Care Med. 2012 Jul;13(4):e234-9. doi: 10.1097/PCC.0b013e318238b162.
Selected optimal respiratory cycles should allow calculation of respiratory mechanic parameters focusing on patient-ventilator interaction. New computer software automatically selecting optimal breaths and respiratory mechanics derived from those cycles are evaluated.
Retrospective study.
University level III neonatal intensive care unit.
Ten mins synchronized intermittent mandatory ventilation and assist/control ventilation recordings from ten newborns.
The ventilator provided respiratory mechanic data (ventilator respiratory cycles) every 10 secs. Pressure, flow, and volume waves and pressure-volume, pressure-flow, and volume-flow loops were reconstructed from continuous pressure-volume recordings. Visual assessment determined assisted leak-free optimal respiratory cycles (selected respiratory cycles). New software graded the quality of cycles (automated respiratory cycles). Respiratory mechanic values were derived from both sets of optimal cycles. We evaluated quality selection and compared mean values and their variability according to ventilatory mode and respiratory mechanic provenance. To assess discriminating power, all 45 "t" values obtained from interpatient comparisons were compared for each respiratory mechanic parameter.
A total of 11,724 breaths are evaluated. Automated respiratory cycle/selected respiratory cycle selections agreement is high: 88% of maximal κ with linear weighting. Specificity and positive predictive values are 0.98 and 0.96, respectively. Averaged values are similar between automated respiratory cycle and ventilator respiratory cycle. C20/C alone is markedly decreased in automated respiratory cycle (1.27 ± 0.37 vs. 1.81 ± 0.67). Tidal volume apparent similarity disappears in assist/control: automated respiratory cycle tidal volume (4.8 ± 1.0 mL/kg) is significantly lower than for ventilator respiratory cycle (5.6 ± 1.8 mL/kg). Coefficients of variation decrease for all automated respiratory cycle parameters in all infants. "t" values from ventilator respiratory cycle data are two to three times higher than ventilator respiratory cycles.
Automated selection is highly specific. Automated respiratory cycle reflects most the interaction of both ventilator and patient. Improving discriminating power of ventilator monitoring will likely help in assessing disease status and following trends. Averaged parameters derived from automated respiratory cycles are more precise and could be displayed by ventilators to improve real-time fine tuning of ventilator settings.
选择最佳呼吸循环可以计算呼吸力学参数，重点关注患者-呼吸机的相互作用。评估一种新的计算机软件，该软件可以自动选择最佳呼吸，并从这些循环中获得呼吸力学参数。
回顾性研究。
大学三级新生儿重症监护病房。
十名新生儿十分钟同步间歇指令通气和辅助/控制通气记录。
呼吸机每 10 秒提供一次呼吸力学数据（呼吸机呼吸循环）。压力、流量和体积波以及压力-体积、压力-流量和体积-流量环从连续压力-体积记录中重建。视觉评估确定了辅助无泄漏的最佳呼吸循环（选择的呼吸循环）。新软件对循环质量进行分级（自动呼吸循环）。从两组最佳循环中得出呼吸力学值。我们评估了质量选择，并根据通气模式和呼吸力学来源比较了平均值及其变异性。为了评估鉴别能力，比较了每个呼吸力学参数的所有 45 个“t”值的患者间差异。
共评估了 11724 次呼吸。自动呼吸循环/选择的呼吸循环选择的一致性很高：最大κ值的线性加权为 0.88。特异性和阳性预测值分别为 0.98 和 0.96。自动呼吸循环和呼吸机呼吸循环的平均值相似。单独的 C20/C 在自动呼吸循环中明显降低（1.27±0.37 比 1.81±0.67）。辅助/控制通气中潮气量的相似性消失：自动呼吸循环的潮气量（4.8±1.0 毫升/公斤）明显低于呼吸机呼吸循环（5.6±1.8 毫升/公斤）。所有婴儿的所有自动呼吸循环参数的变异系数均降低。呼吸机呼吸循环数据的“t”值是呼吸机呼吸循环的两倍到三倍。
自动选择具有高度特异性。自动呼吸循环反映了呼吸机和患者相互作用的大部分情况。提高呼吸机监测的鉴别能力可能有助于评估疾病状态和监测趋势。从自动呼吸循环中得出的平均值更精确，并且可以由呼吸机显示，以改善呼吸机设置的实时微调。
