Fig. 4:
From: Accounting for treatment use when validating a prognostic model: a simulation study
Calibration curves calculated in a treated validation set, following different approaches to account for the effects of treatment, in the presence of an unmeasured predictor (U) associated with both the outcome and the probability of receiving treatment. Scenario 13: U has a weak association with the outcome (log(OR) = 1); scenario 14: U has a moderate association with the outcome (log(OR) = 2); scenario 15: U has a strong association with the outcome (log(OR) = 4). Plots were based on sets of 1 million individuals.