Fig. 2
From: Dynamic prediction based on variability of a longitudinal biomarker
Model 4 (M4) allowed each individual to have it’s own random error term, to account for varying amounts of variability in TAC between patients. The distribution of the individual’s standard deviations (SD) is shown here as a histogram. The red vertical line represents the common standard deviation taken from M1, which forced each individual to have a common residual error