• Enable optimal disposition decision-making by emergency department (ED) clinicians through provision of data-driven clinical decision support (CDS) at the point of care. Through the application of supervised machine learning methods to electronic health record (EHR) data, we will develop models that predict clinical deterioration and advanced care requirements at the individual patient level. Predictions will be translated to actionable risk-driven disposition recommendations, embedded within existing clinical workflow.

• Enable optimal resource allocation across a hospital system through provision of data-driven decision support to a central Capacity Command Center. We will utilize mathematical models of regional and state-level hospital systems to achieve optimal load-balancing of COVID-19 patients. The optimization models will consider various criteria including the current and forecasted available capacity of each hospital, the transfer time between hospitals, existing relationships between hospitals, availability of required resources, and the distribution of inpatient length-of-stay.

• Forecast COVID-19 cases, deaths, and hospitalizations and assess policy choices regarding both pharmaceutical and non-pharmaceutical interventions. Using both an age-weighted compartmental model and an individual-based model of SARS-CoV-2, we will assess short-term trends in the number of cases, deaths, and hospitalizations at the state and/or county level for the United States. In addition, we will assess current policy options regarding different types of non-pharmaceutical interventions and the policy implications of pharmaceutical interventions, including vaccines and therapies, as they become available.

Status