|
Abstract:
|
Dose-response experiments are routinely conducted to examine response rates at various dose levels of interest. Due to side effects, lack of efficacy or other reasons, not all responses are observed in practice. Consequently, our final design can be inefficient. Baek et al. (2006) proposed to construct single and multiple-objective optimal designs for a binary dose response study when potentially missing observations are incorporated at the onset of the study. In this paper, we compared the performance between commonly used designs in clinical trials and the proposed optimal designs accounting for potential missing observations in terms of the precision of related parameter estimations, for example, the estimation of certain dose levels or the underlying dose response model parameters.
|
