Locally optimal designs for some dose–response models with continuous endpoints

ABSTRACT

We consider the problem of constructing static (or non sequential), approximate optimal designs for a class of dose–response models with continuous outcomes. We obtain conditions for a design being D-optimal or c-optimal. The designs are locally optimal in that they depend on the model parameters. The efficiency studies show that these designs have high efficiency when the mis-specification of the initial values of model parameters is not severe. A case study indicates that using an optimal design may result in a significant saving of resources.

KEYWORDS:

• c-Optimality
• D-optimality
• dose–response
• efficiency
• toxicology.

MATHEMATICS SUBJECT CLASSIFICATION:

• Primary 62K05
• 62G35
• Secondary 62G07

Additional information

Funding

The research of Zhide Fang is supported by grant 1 U54 GM104940 from the National Institute of General Medical Sciences of the National Institutes of Health, which funds the Louisiana Clinical and Translational Science Center of Pennington Biomedical Research Center.