Critically ill patients admitted to the Intensive Care Unit (ICU) are at the risk of developing sepsis leading to death, each hour of delayed recognition or treatment can contribute to significantly increasing the likelihood of death. If this application is funded, we will explore sensor-generated continuous physiological data streams to extract ‘physiomarkers’ that predict the onset of sepsis, aid in informing intervention futility (volume responsiveness) and support the discovery of novel sepsis sub-types. By discovering and modeling the mechanisms by which such physiomarkers predict sepsis, we can proactively recognize specific high-risk patients, appropriate resources at the right time to avoid significant morbidity and mortality, and begin treatments that improve outcomes in this critically ill and vulnerable population.

 In the parent award, we investigate the physiological mechanisms that contribute to increased likelihood sepsis among patients admitted to the ICU. In this supplement we evaluate possible causes of uncertainty by utilizing conformal predictions and hypothesis tests. We will derive a novel framework which we term EQUITI that can be used to characterize the degree to which the model is uncertain, due to the influence of bias in the data. Furthermore, this supplement will also contribute workforce and skills development aids that are proposed to be used to improve clinicians and healthcare professionals understanding of ambiguities in model estimated output. Finally, these aids will also help health professionals better understand how these uncertainties can be identified and used to improve individual and team situational awareness.

"Patient-Focused Collaborative Repository Uniting Standards for Equitable AI," seeking to develop a publicly available, high-resolution, labeled dataset of unprecedented #ICU demographic and treatment diversity, uniting 19+ centers.