An effective method of construction of a linear\nestimator of AUC in the finite interval, optimal in the\nminimax sense, is developed and demonstrated for five PK\nmodels. The models may be given as an explicit C(t) relationship\nor defined by differential equations. For high\nvariability and rich sampling the optimal method is only\nmoderately advantageous over optimal trapezoid or standard\nnumerical approaches (Gauss-Legendre or Clenshaw-\nCurtis quadratures). The difference between the optimal\nestimator and other methods becomes more pronounced\nwith a decrease in sample size or decrease in the variability.\nThe described estimation method may appear useful\nin development of limited-sampling strategies for AUC\ndetermination, as an alternative to the widely used\nregression-based approach. It is indicated that many alternative\napproaches are also possible.
Loading....