Skip to main content
Fig. 3 | BMC Medical Informatics and Decision Making

Fig. 3

From: Learning patient-level prediction models across multiple healthcare databases: evaluation of ensembles for increasing model transportability

Fig. 3

Box plots of calibration-in-the-large (observed risk—mean predicted risk) values for each non-ensemble (Level 1 model) and ensemble (Level 2 model) when externally validated. The rows represent the external database (the database that was excluded from the model/ensemble development) that was used to fairly evaluate the models/ensembles. The x-axis represents the model/ensemble technique. Box plots centered around 0 with a small range indicate excellent external calibration performance. The dashed vertical lines separate the non-ensembles, the fusion ensembles, the mixture of expert ensembles and the stacking ensembles

Back to article page