GAINING CONSENSUS ON EXPERT RULE STATEMENTS FOR A DIGITAL TWIN PATIENT MODEL IN THE ICU USING A DELPHI METHOD

ABSTRACT

SESSION TITLE Optimizing Resources in the ICU SESSION TYPE Original Investigations PRESENTED ON 10/16/2022 1030 am - 1130 am PURPOSE: The COVID-19 pandemic has exposed worldwide heterogeneity in the application of fundamental critical care principles and best practices. New methods and strategies to facilitate timely and accurate interventions are needed. If built on a robust foundation of physiologic principles, a virtual critically ill patient (aka digital twin) could better inform decision making in critical care. When used in clinical practice, a digital twin may allow bedside providers to preview how organ systems interact to cause a clinical effect, providing the opportunity to test the effects of various interventions virtually, without exposing an actual patient to potential harm. Building on our previous work with a digital twin model of critically ill patients with sepsis, this current project focuses specifically on the respiratory system. METHODS: We assembled a modified Delphi panel of 36 international critical care experts. We modeled elements of respiratory system pathophysiology using directed acyclic graphs (DAG) and derived several statements describing associated ICU clinical processes. Panelists participated in three Delphi rounds to gauge agreement on 71 final statements using a 6-point Likert scale. Agreement was defined as >80% selection of a 5 (“agree”) or 6 (“strongly agree”). RESULTS: The first Delphi round included statements of pulmonary physiology affecting critically ill patients, eg pulmonary edema, hypoxemic and hypercapnic respiratory failure, shock, acute respiratory distress syndrome (ARDS), airway obstruction, restrictive lung disease, and ventilation-perfusion mismatch. Agreement was achieved on 60 (84.5%) of expert statements after completion of two rounds. After partial completion of the third round, agreement increased to 62 (87%). Statements with the most agreement included the physiology and management of airway obstruction decreasing alveolar ventilation and the effects of alveolar infiltrates on ventilation-perfusion matching. Lowest agreement was noted for the statements describing the interaction between shock and hypoxemic respiratory failure due to increased oxygen consumption and ARDS increasing dead space. CONCLUSIONS: An international cohort of critical care experts reached 87% agreement on our rule statements for respiratory system pathophysiology. The Delphi approach appears to be an effective way to refine content for our digital twin model. CLINICAL IMPLICATIONS Expert consensus can be used to strengthen the respiratory physiology statements used to direct the ICU digital twin patient model. With a digital twin based on refined respiratory physiology statements, bedside providers may preview how organ systems interact to cause a clinical effect without exposing an actual patient to various interventions. DISCLOSURES No relevant relationships by Ognjen Gajic, value=Royalty Removed 06/06/2022 by Ognjen Gajic No relevant relationships by Amos Lal No relevant relationships by John Litell No relevant relationships by Amy Montgomery