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

Table 5 Parameter estimates in landmarking supermodel using two-types of dynamic pseudo observations (DPOs)

	DPOs based on AJ estimator (eq.(5))				DPOs based on KM estimator (eq.(6))
	1 recurrence		2 or more recurrences		1 recurrence		2 or more recurrences
	estimate	robust s.e.	estimate	robust s.e.	estimate	robust s.e.	estimate	robust s.e.
Intercept	−1.22	1.99	1.30	4.27	−1.06	1.97	−0.13	4.23
Time (year)
s	2.19	3.28	−6.24	6.28	1.92	3.24	−3.98	6.15
s²	−1.53	1.64	2.42	2.69	−1.41	1.62	1.38	2.61
s³	0.22	0.24	−0.32	0.34	0.21	0.24	−0.19	0.33
The number of recurrences (1 or more / 0)
Intercept	1.92	2.12	0.49	4.72	1.84	2.09	1.51	4.54
s	−2.87	3.26	3.36	7.01	−2.70	3.21	1.78	6.73
s²	1.65	1.55	−2.26	2.99	1.58	1.53	−1.55	2.87
s³	−0.24	0.22	0.38	0.38	−0.23	0.22	0.29	0.37
Multiple tumors / single tumor
Intercept	4.46	2.09	−2.27	4.32	4.31	2.05	−1.30	4.44
s	−7.87	3.21	3.16	6.64	−7.57	3.15	1.38	6.77
s²	3.86	1.50	−0.91	2.89	3.71	1.48	0.01	2.91
s³	−0.53	0.21	0.09	0.37	−0.51	0.21	−0.04	0.36
The length of tumor (> 2 cm / ≤2 cm)
Intercept	−0.10	1.87	−1.19	4.07	−0.09	1.89	−0.66	4.00
s	0.31	2.88	4.08	6.21	0.30	2.91	3.29	6.04
s²	−0.21	1.34	−2.05	2.70	−0.19	1.37	−1.74	2.58
s³	0.03	0.19	0.29	0.35	0.02	0.20	0.26	0.33

ISSN: 1471-2288