With an associated mortality of 35%, Acute Respiratory Distress Syndrome (ARDS) contributes to the morbidity and mortality seen in many acute conditions such as pneumonia, influenza, sepsis, trauma and aspiration. Yet studies consistently show that ARDS is under-recognized in up to 40% of patients with generally low and variable adoption of practices shown to improve mortality in ARDS. The overall goal of this proposed project is to improve care delivered to ARDS patients by promoting accurate identification of patients with ARDS and increasing adoption of evidence-based interventions shown to improve outcomes in ARDS. With the recent publication of a multi-professional Clinical Practice Guideline (CPG) on mechanical ventilation in ARDS, this proposal specifically aims: 1) To increase identification of ARDS patients with an automated electronic tool (ARDS Sniffer 2.0) powered by the predictive analytics of machine learning algorithms and natural language processing. 2) To translate the recently published multi-professional Clinical Practice Guideline (CPG) on mechanical ventilation in ARDS into an evidence-based, context-appropriate, Electronic Clinical decision support system in ARDS (ECARDS) for the management of ARDS. This process will be informed by an expert panel from the CPG committee as well as a semi-qualitative analysis of the cognitive decision making process, data needs and workflow of users in a series of Think-Aloud studies and focus group interviews. ARDS Sniffer 2.0, ECARDS and an ARDS Dashboard to show near real-time data on the incidence of ARDS and rate of utilization of best care will then be incorporated within the hospital EMR for useability testing with another series of Think-Aloud and semi-qualitative studies. 3) To evaluate the effectiveness of ARDS Sniffer 2.0, ARDS Dashboard, and ECARDS in a stepped-wedge, cluster randomized control trial of 3 hospitals within the Montefiore Healthcare System to increase utilization of recommended interventions in ARDS and improve outcomes. 4) To promote the dissemination of the clinical decision support system through our partner professional organizations of American Thoracic Society, Society of Critical Care Medicine, and American Association of Respiratory Care and Program for Emergency Preparedness. Because of the high pressure, high acuity, and rapidly changing status of the acutely ill ARDS patient, clinical decision making often defaults to the automated, pattern-driven cognitive processes which are particularly prone to unconscious biases and errors and cognitive overload from large volumes of shifting data. As such, this proposal is the ideal research demonstration project to develop a new clinical decision support tool that will harness the power of big data and predictive analytics to identify ARDS in the hospital and deliver an evidenced-based, expert and user informed clinical decision support tool to clinicians at the right point during their clinical interaction and workflow with these patients in order to promote the right interventions to the right patients at the right time.
Even as Acute Respiratory Distress Syndrome constitutes the most severe form of respiratory failure in the hospital with an associated mortality of 35%, it is consistently under-recognized and proven treatments are underutilized. This proposal aims to harness the power of big data in the hospital information system to identify patients with ARDS and prompt clinicians with the evidence-based best management of this condition. By developing a computerized decision support system whose design is informed by the experts and the clinicians themselves, this proposal aims to bring the right care to the right patient at the right time to save lives and promote recovery from this devastating condition.
