Objective: To describe and evaluate a new weaning and decision support system for mechanical ventilation.
Background: FLEX is a computerized weaning and decision support system for mechanical ventilation that unlike previous rule-based systems derives many of its rules on the basis of the conditions of individual patients. This system can be used in a wide range of ventilatory modes as well as automatic control of weaning. It incorporates the features of the patented ventilatory mode known as Adaptive Support Ventilation (ASV) along with other new features for control of weaning, and control of patient's oxygenation by adjustment of PEEP and the fraction of inspired oxygen.
Methods: Ventilator data was collected for 10 patients in medical/surgical ICU at baseline and about 24 hours later. Required data fields for each patient for these two time points were also entered into the FLEX program. Comparison of clinical data and FLEX recommendations were made with regard to minute ventilation, alarms, weaning institution and other variables.
Results: At baseline, 7 patients were being treated with AC, the remainder with IMV/PS. There was good agreement between the measured and recommended minute ventilations; variances were seen in some patients being treated with permissive hypercapnea and those with evidence of high oxygen needs or other metabolic derangements. At 24 hours, there was improved correlation between measured minute ventilation and that recommended by FLEX, suggesting that clinical adjustments were in-line with Flex recommendations over time. Furthermore, FLEX made recommendations with regard to FIO(2) and PEEP that would potentially diminish the risk of oxygen toxicity, hypoxemia, and barotrauma in selected patients. FLEX has also been implemented as a closed loop system in an initial set up.
Conclusion: A new weaning and decision support system for mechanical ventilation is presented. The recommendations made by the system were found to be in line with clinical determinations. Further refinements in the FLEX predictions can be easily made by including inputs which represent permissive hypercapnea or increased metabolic demand for selected patients.