Fig. 3From: Framework for personalized prediction of treatment response in relapsing remitting multiple sclerosisCalibration. Calibration of the relapse model (top) and CDP model (bottom) using equally-populated bins, when considering all DMTs together (left) and when considering each DMT separately (right). The predictions are split into 20 equally-populated bins from zero to four relapses, for the relapse model, and from zero to one, for the CDP model. Different shades of gray highlight a different population sizeBack to article page