Table 1. Description of prior distributions and hyper-parameters for model parameters.
Model | Parameter | Prior Distribution | Description |
HMM1,2 | μ1 | Normal(0,0.01)* | Mean state 1 |
HMM1,2 | μ2 | Normal(0,0.01)* | Mean state 2 |
HMM1,2 | PInit | Dirichlet(0.5,0.5) | Initial Probability |
HMM1,2 | P | Beta(0.5,0.5) | Probability transition matrix |
HMM1,2 | Y | Poisson() | Observed count data |
HMM2 | X | Normal(0,0.001)* | Covariate coefficients |
*Parameterized as mean and precision (1/variance, as in WinBUGS). For disease-level models, a Normal(0,10) prior was used to accommodate very small expected counts.