A Proposed Approach for Joint Modeling of the Longitudinal and Time-To-Event Data in Heterogeneous Populations: An Application to HIV/AIDS’s Disease
Table 3
Comparison of the estimation of parameters in the three approaches for different sample sizes, .
Models
Class
Parameter
True value
Bias
SE
CP
Bias
SE
CP
Bias
SE
CP
JLCM
Class 1
Intercept
0.5
0.144
0.231
0.631
0.122
0.210
0.629
0.103
0.159
0.711
Class 2
Intercept
−0.5
0.176
0.332
0.678
0.142
0.249
0.673
0.108
0.167
0.785
PA
Class 1
Intercept
0.5
0.024
0.208
0.941
0.019
0.171
0.947
0.012
0.142
0.976
0.5
−0.017
0.177
0.950
−0.012
0.153
0.951
−0.002
0.149
0.952
Class 2
Intercept
−0.5
0.032
0.201
0.933
0.024
0.183
0.949
0.013
0.155
0.982
0.5
−0.019
0.148
0.915
−0.015
0.137
0.951
−0.003
0.121
0.987
SA
Class 1
Intercept
0.5
0.101
0.288
0.728
0.091
0.179
0.813
0.084
0.157
0.817
0.5
−0.219
0.321
0.669
−0.210
0.276
0.671
−0.195
0.225
0.698
Class 2
Intercept
−0.5
0.098
0.259
0.709
0.086
0.219
0.691
0.082
0.189
0.693
0.5
−0.234
0.277
0.649
−0.210
0.228
0.718
−0.198
0.207
0.708
JLCM: joint latent class models; PA: proposed approach; SA: separate approach; SE: standard error; CP: coverage probability of the 95% confidence interval.