Evaluación e impacto del uso secundario de los datos del sistema de información clínica para la identificación de parámetros críticos, desarrollo de indicadores y modelos predicitivos en ventilación mecánica

Identifier:  TDX:4531

Handle: https://hdl.handle.net/20.500.11797/TDX4531

Authors:  Manrique Moreno, Sara

Abstract:
Many invasive mechanical ventilation (IMV) parameters have been suggested as harmful. However, most of the published studies have been conducted with snapshots of IMV. But IMV is continuous, and VILI can occur at any moment during its application, so these studies do not reflect the reality of ventilated critically ill patients. In this context, this doctoral thesis uses continuously extracted data throughout the entire duration of IMV from the clinical information system (CIS) to identify critical parameters and develop new quality indicators and predictive models in IMV. For this purpose, various univariate and multivariate analyses have been conducted to detect ventilatory parameters related to clinical outcomes in patients admitted to and ventilated in the ICU. The two new proposed indicators about time spent with an appropriate tidal volume (Vt) and the total number of hours with mechanical power (MP) > 18 J/min were analyzed as independent variables. Additionally, a predictive model for MP has been developed to alert clinicians when safe limits are about to be exceeded to prevent damage before it occurs. It has been shown that adherence to clinical practice guidelines for protective IMV remains low and that patients who spend more than 80% of the time with Vt < 8ml/KgIBW have fewer days on IMV and a shorter ICU stay, although there are no significant differences in mortality. However, the total number of hours with MP > 18 J/min is significantly associated with increased mortality, particularly in hypoxemic patients. This doctoral thesis has also demonstrated that developing and implementing a predictive artificial intelligence model for MP is feasible and could reduce VILI