Fig. 1From: Informative missingness in electronic health record systems: the curse of knowingImpact of differences in missing data mechanisms when applying a prediction model that was derived under informative missingness using different approaches to handle missing data. Plots are based on 500 randomly sampled observations out of a dataset of 20000 observations. Predicted probabilities are jittered for visual clarity. See main text for a description of the scenarios and details about the methodsBack to article page