Fig. 2From: Validation of a machine learning algorithm for early severe sepsis prediction: a retrospective study predicting severe sepsis up to 48 h in advance using a diverse dataset from 461 US hospitalsAUROC over time. Depicts performance of the MLA in predicting the onset of severe sepsis at 0, 4, 6, 12, 24 and 48 h before severe sepsis onset. “Training Set” results were derived from the DAD, “Testing Set” results were derived from the hold out data from the DAD, and the “External Validation Set” was derived from the independent CHHDBack to article page