Fig. 1From: A machine learning approach using endpoint adjudication committee labels for the identification of sepsis predictors at the emergency departmentReceiver operator characteristic (ROC) and precision recall (PRC) curves of the three models with the best performance of each model configuration in the double loop cross validation (DLCV) scheme. Three model configurations are shown; base (logistic regression on demographic and vital data), extended L1 (lasso on demographic, vital, laboratory and sapphire data) and extended RF (random forest on demographic, vital, laboratory and sapphire data). For each of the three model configurations, ten models were trained on different data splits in the tenfold DLCV scheme. During each iteration, predictions were computed with on both training and test folds. Both ROC and PRC curves were drawn with the aggregated train and test data over 10 folds for each model configurationBack to article page