In medical drug therapy, efficient dosage strategies are needed to maintain target drug concentrations. The relationship between the concentration of a drug and the dosages is often described by compartment models in which the parameters are unknown, although prior knowledge may be available and can be updated after blood samples are taken during the therapy. Currently MAP (maximum a posteriori) Bayesian is the most often used control strategy in this setting. We show by simulation in a one-compartment context that the performance of the MAP Bayesian strategy depends on the assumptions in prior distribution of the parameters as well as the cost function. We propose an alternative control strategy, VU, that outperforms and is more robust than the MAP Bayesian strategy in a variety of problem settings.
展开▼
