Olliver S, Davis G M, Hatzakis G E
Department of Pediatrics, Montreal Children's Hospital, McGill University Health Center, Canada.
AMIA Annu Symp Proc. 2003;2003:499-503.
We have previously developed a fuzzy logic controller for weaning adults with chronic obstructive pulmonary disease using pressure support ventilation (PSV). We used the core of our fuzzy logic-based weaning platform and further developed parametrizable components for weaning newborns of differing body size and disease-state. The controller was validated on neonates recovering from congenital heart disease (CHD) while receiving synchronous intermittent mandatory ventilation (SIMV). We wished to compare the efficacy of this controller versus the bedside weaning protocol in children with respiratory syncytial virus pneumonitis/bronchiolitis (RSV) in the pediatric intensive care unit (PICU). The fuzzy controller evaluated the "current" and "trend" weaning status of the newborn to quantitatively determine the change in the SIMV integrated ventilatory setting. For the "current" status it used heart rate (HR), respiratory rate (RR), tidal volume (VT) and oxygen saturation (SaO2), while for the "trend" status the differences of deltaRR/ deltat, deltaHR/ deltat, and deltaSaO2/ deltat recorded between two subsequent time points were utilized. The enumerated vital signs were fuzzified and then probability levels of occurrence were assigned. Individualized "golden" goals for SaO2 were set for each newborn. We retrospectively assessed the charts of 19 newborns, 113+/-128 days old, 5,546+/-2,321 gr body weight, weaning for 99+/-46 days, at 2-hour intervals. The SIMV levels proposed by the fuzzy controller were matched to those levels actually applied. In 60% of the time both values coincided. For the remaining 40%, the controller was more aggressive suggesting lower values of SIMV than the applied ones. The Area under the SIMV curves over time was 1,969+/-1,044 for the applied vs 1,886+/-978 for the suggested levels, respectively. The fuzzy controller adjusted for body size and disease-pattern can approximate the actual weaning course of newborns with RSV.
我们之前开发了一种模糊逻辑控制器，用于对患有慢性阻塞性肺疾病的成人进行撤机，采用压力支持通气（PSV）模式。我们利用基于模糊逻辑的撤机平台的核心，并进一步开发了可参数化组件，用于对不同体型和疾病状态的新生儿进行撤机。该控制器在接受同步间歇指令通气（SIMV）的先天性心脏病（CHD）恢复期新生儿身上进行了验证。我们希望在儿科重症监护病房（PICU）中，比较该控制器与床边撤机方案对患有呼吸道合胞病毒肺炎/细支气管炎（RSV）儿童的疗效。模糊控制器评估新生儿的“当前”和“趋势”撤机状态，以定量确定SIMV综合通气设置的变化。对于“当前”状态，它使用心率（HR）、呼吸频率（RR）、潮气量（VT）和血氧饱和度（SaO2），而对于“趋势”状态，则利用两个连续时间点记录的deltaRR/deltat、deltaHR/deltat和deltaSaO2/deltat的差值。将列举的生命体征进行模糊化处理，然后分配发生的概率水平。为每个新生儿设定个性化的SaO2“黄金”目标。我们回顾性评估了19例新生儿的病历，这些新生儿年龄为113±128天，体重5546±2321克，撤机时间为99±46天，每隔2小时记录一次。将模糊控制器提出的SIMV水平与实际应用的水平进行匹配。在60%的时间里，两个值是一致的。在其余40%的时间里，控制器更激进，建议的SIMV值低于实际应用的值。随着时间推移，实际应用的SIMV曲线下面积为1969±1044，而建议水平的SIMV曲线下面积为1886±978。针对体型和疾病模式进行调整的模糊控制器可以近似患有RSV的新生儿的实际撤机过程。
