Chapter 8 - Monitoring Each Patient Optimally: When to Obtain the Best Samples for Therapeutic Drug Monitoring

Abstract

In this chapter, we discuss the most informative times to obtain data for therapeutic drug monitoring. One well-known class of designs is called D-optimal, which is designed mainly for parametric models and is oriented to obtaining the best model parameter estimates. There is also a new approach, that of multiple model optimal design, which is especially well suited for nonparametric (NP) models. This design finds the sampling times that minimize the risk of mistaking one NP support point for another. It consistently outperforms the various D-optimal designs. It also can be oriented to a clinical task to be performed, such as achieving a desired target goal or an area under the curve, with maximal precision. It represents a totally new approach to the clinical problem of optimizing learning about the patient while treating, by doing so with regard to optimizing the specific and practical clinical task to be achieved by the dosage regimen.