This application takes as input patient characteristics/dosing schedule and presents the population-level expected response as well as the individual-level expected response. The individual-level responses are rendered with empirical Bayesian estimates (EBES), so they require an observed blood level draw. The Bayesian process is based on the minimization of the likelihood with respect to the individual random effects [1]. The tool uses two models: Taylor et al. [2] and Blackman et al. [3]. This is based on the findings in Blackman et al., a validation study of high-dose methotrexate (HD-MTX) models. The study found that Blackman et al. was superior at high, early concentrations where toxicities are most likely, and that Taylor et al. was superior at low, later concentrations (especially with the use of EBEs. The unique ability of this application to provide predictions from both models enables the clinician to make the best decision at any time during a patient's concentration-time curve.
The models implemented in this application are all available in the scientific literature. The calculation of the EBEs for individual random effects is outlined in Kang 2012 [1] and the code to calculate the EBEs is adapted from the open source TDM software mrgsolve.