LAUSR

Briefly, the Bayesian approach can be described as follows. The simulated adherent and nonadherent scenarios were identical with that used in percentile method (i.e., 4 and 8 scenarios for first-dose DOT and non-DOT criteria). Two-thousand individual DEAQ concentration-time profiles were generated for each scenario, based on the proposed population PK model. All simulated concentrations at time t from all scenarios (2,000 individual DEAQ concentrations per scenario) were pooled together and 20 concentration bins from <5% to >95% percentiles (every 5percentage unit) based on the distribution of pooled concentrations were generated. The conditional probability of a concentration for a given scenario was calculated in each of these 20 bins. Incorporation of the conditional probability and the equiprobable prior probability, the posterior probability for each scenario at a given concentration was calculated with the Bayesian formula (equation 1 and 2). The most plausible scenario of a given concentration at time t, is the one with the largest posterior probability among all scenarios.