Bayesian inference in pharmacokinetics/pharmacodynamics using regenerative methods

To obtain independent and identically distributed subchains that allow application of regenerative simulation analysis method, a splitting technique was applied for Independence Metropolis-Hastings (IMH) chains and a hybrid sampler, Rejection Independence Metropolis-Hastings (RIMH) algorithm, in pharmacokinetic/pharmacodynamic parameter estimation. The confidence intervals associated with posterior quantities were evaluated using an extensive coverage study with a model for the anticancer drug Teniposide. For this example, both methods demonstrated coverage of 95% confidence intervals for posterior means and selected quantiles of model parameters and predicted drug concentrations close to their nominal values.