“…The simulated 8-patient PK datasets provided a good visual match to the trial data from McCarthy et al [14] (Figure S1). The PK model was incorporated into a Bayesian hierarchical framework, and fitted to each of the 1000 simulated datasets, restricting data to the cipargamin concentrations which correspond to the sampling times of the original Phase 2 trial (1,2,3,4,6,8,12,16,24,36,48,72,96 and 120 hours post-treatment), and the posterior median estimate of each population PK parameter obtained. To evaluate how accurately this model can estimate PK parameters, we calculated the difference (absolute and relative bias) between the posterior median estimate of the population-level PK parameter, and the value used to simulate the data (i.e., the 'true' value).…”