Figure 2 . Interpolation resolves discrepancy between the predicted value distribution and the true distribution of expected future time when using ExtraTreeRegressor as the base learner. a. Using a validation set to interpolate real-world distribution. b. Interpolation resolves the discrepancy between predicted values and the gold standard rwTTD curve. c. Comparison between the distribution of prediction values and gold standard rwTTD future time. d. Histogram of error rates at different evaluation maximal dates. e. Absolute error dates when 50% of the population is terminated.