Skip to main content

Table 4 Simulation results in the presence of a terminal event

From: Dynamic prediction of repeated events data based on landmarking model: application to colorectal liver metastases data

Scenario DPOs based on AJ estimatora
u i λ 01 λ 02 λ 0 D λ c no events 1 event 2 events a terminal event
Absolute biasb
 1 1 1 0.3 0.5 0.0003 − 0.00002 0.0006 −0.0008
 1 1 2 0.3 0.5 0.0003 −0.0004 0.0010 −0.0008
 Γ (0.5, 0.5) 1 1 0.3 0.5 0.0011 −0.0002 − 0.0011 0.0002
 Γ (0.5, 0.5) 1 2 0.3 0.5 0.0011 −0.0002 − 0.0010 0.0002
 1 1 1 0.3 2 −0.0025 −0.0017 0.0040 0.0002
 1 1 2 0.3 2 −0.0057 0.0088 −0.0032 0.0001
 Γ (0.5, 0.5) 1 1 0.3 2 −0.0081 0.0086 0.0055 −0.0060
 Γ (0.5, 0.5) 1 2 0.3 2 −0.0094 0.0137 0.0010 −0.0053
Root Mean Squared Error (RMSE)
 1 1 1 0.3 0.5 0.0264 0.0316 0.0241 0.0245
 1 1 2 0.3 0.5 0.0264 0.0270 0.0281 0.0245
 Γ (0.5, 0.5) 1 1 0.3 0.5 0.0261 0.0238 0.0207 0.0211
 Γ (0.5, 0.5) 1 2 0.3 0.5 0.0261 0.0214 0.0225 0.0209
 1 1 1 0.3 2 0.0724 0.0836 0.0697 0.0681
 1 1 2 0.3 2 0.0704 0.0749 0.0783 0.0687
 Γ (0.5, 0.5) 1 1 0.3 2 0.0703 0.0644 0.0526 0.0562
 Γ (0.5, 0.5) 1 2 0.3 2 0.0701 0.0609 0.0620 0.0575
  1. a Proposed in eq.(5) and eq.(7)
  2. b Mean differences between true values and dynamic predicted values; True values are empirical probabilities of event numbers calculated using potential event times. Dynamic predicted values are expectations of proposed DPOs