ABSTRACT
FIGURE 7.10: ROC curves at time t = 4 and three options for t under the simple prediction rule (based on the joint model tted to the Liver Cirrhosis dataset).
using a for-loop. The following code illustrates this based on the plcbData data frame:
At each iteration, one extra row of plcbData is taken into account, and thus we obtain the estimates of sensitivity and specicity at t = 0, 0.25, 1, 3, and 4 years. The corresponding time-varying ROC curves starting from t = 0:25 years, and for the three dierent options of t, are graphically illustrated in Figure 7.11. The R code to produce this graph uses a similar for-loop:
FIGURE 7.11: Time-dependent ROC curves for the liver cirrhosis dataset under the simple prediction rule. The horizontal solid lines denote the sensitivity levels corresponding false positive rate of 20%.
