Modeling the weaning of intensive care unit patients from mechanical ventilation: a review

In the intensive care unit, mechanical ventilation is a life-saving procedure, and as many as 90% of patients require the intervention. For a mechanically ventilated patient, the principal goal of a health care team is to free the patient from mechanical ventilation through weaning as soon as possible. Weaning, however, still is mostly a manual process. To achieve quick and efficient weaning, the process is needs to be automated. The first step toward automating the weaning process is building a precise model of it. The path to achieving this precision in weaning modeling, if at all possible, is laden with challenges such as the use of imprecise terms, lack of evidence, complexities in data representation as well as process specification, and uncertainty in data values as well as their implication in process evaluation. This eventually leads to a lack of universally accepted and followed standards and guidelines. Despite the magnitude of these challenges, various weaning automations have been attempted through mathematical modeling or knowledge-based modeling. Some of these have been available as commercial mechanical ventilator modes since the 1990s. Even though much potential has been demonstrated through clinical trials, their infrequent usage indicates a lack of consensus concerning their applicability.