Abstract:
This research focuses on practical considerations for establishing efficient study design
to estimate target dose of interest in treating hypertensive patients and we consider
optimal design for the estimation of the minimum effective dose to the maximum
treatment effect based on the data collected. We obtained this through sequential
construction of optimal design, using D-optimality criterion.An optimal design of two
drugs (Aldoment and Lisinopril) used in treating high blood pressure was constructed
using D-optimality criterion along with a non-linear regression model. The relative
efficiencies of other designs with respect to the optimal design were determined. Due to
the nonlinearity of the model, we supplied initial values for the model parameters by
using least square parameter estimates before running the algorithm. A 3 by 2 (3x2)
starting design matrix was provided and the algorithm iterates until the stopping
criterionwas satisfied, that is, when the standardized variance p, where p is the number
of parameters in the model. The procedure typically iterates by mixing the current
design with a specially selected point to form a new design, the selected point to
introduce in the iteration is the point that maximizes the weighted variance function.
The result shows that the optimal design is the design whose information matrix has the
highest determinant leading to the design point (-1) and (1) with equal weight of 0.5.
