Auto-PEEP is auto positive end-expiratory pressure due to excessive amounts of alveolar gas produced by sustained recurrent incomplete exhalation. Incomplete exhalation occurs when the exhaled breath never reaches a flow rate of 0 L/min. The objective of this dissertation is to develop an automated detection system of auto-PEEP through incomplete exhalation as revealed by ventilator graphics for mechanically ventilated adults. Auto-PEEP can cause adverse effects if allowed to linger and if not quickly identified. An automated detection system will be instrumental in helping to quickly identify auto-PEEP. A computerized algorithm was developed to detect incomplete exhalation based on the following three parameters:1) Foi, was used to represent the value of the flow at the onset of inhalation, 2) ∆T, was used to represent the value of time difference between onset inhalation to the 0 L/min mark, and 3) slope threshold, a value set for the slope of change of flow over ∆T. Optimum parameters of the algorithm were achieved for Foi = -3 L/min, ∆T = 0.2 s, and slope threshold = 90 L-s/min. A novel data set was introduced to validate the algorithm, yielding no significant difference in true positive rates (t = 1.5, df = 12.402, p-value = 0.1408) and false positive rates (t = 1.9, df = 16.765, p-value = 0.0725) as outcomes for two-tailed t-tests comparing the novel and old data set. To determine the relationship between auto-PEEP and detection of sustained incomplete exhalation, a correlation of a linear model of the novel data set between auto-PEEP and the percentage of incomplete exhalation detection out of the existing breaths (index) was investigated. A linear model should interpret the index value that corresponds to significant auto-PEEP presence; unfortunately, no significant linear model was found between incomplete exhalation index and auto-PEEP (F1,62 = 1.67, p-value = 0.2010). However, there was a relationship between the intrinsic PEEP values and the incomplete exhalation index as functions of time. The automated detection algorithm produced by this work provides a non-invasive method of automatically detecting auto-PEEP.
| Identifer | oai:union.ndltd.org:vcu.edu/oai:scholarscompass.vcu.edu:etd-1533 |
| Date | 01 January 2013 |
| Creators | Arief, Nyimas |
| Publisher | VCU Scholars Compass |
| Source Sets | Virginia Commonwealth University |
| Detected Language | English |
| Type | text |
| Format | application/pdf |
| Source | Theses and Dissertations |
| Rights | © The Author |
Page generated in 0.002 seconds