Alliance for Pandemic Preparedness

February 17, 2021

Real-Time Prediction of COVID-19 Related Mortality Using Electronic Health Records

Category:

Topic:

Keywords (Tags): ,

The COVID-19 early warning system (CovEWS), a risk assessment system tool for real-time forecasting of COVID-19 related mortality risk, was found to predict mortality up to 192 hours prior to mortality events in patients hospitalized with COVID-19. The model was constructed from a variety of clinical and biomarker inputs and at a sensitivity of greater than 95%, the specificity was 78.8% at 1 hour before a predicted mortality event and 69.4% at 192 hours before a predicted mortality event, comparing favorably to several other risk stratification tools. The system was developed using electronic health record data from a cohort of 66,430 patients seen at over 69 healthcare institutions. 

Schwab et al. (Feb 2021). Real-Time Prediction of COVID-19 Related Mortality Using Electronic Health Records. Nature Communications. https://doi.org/10.1038/s41467-020-20816-7