OPTIMAL DESIGN FOR THE ELISA TYPE OF IMMUNOASSAYS: CLASSICAL CALIBRATION VERSUS PARALLEL FITTING

摘要

Immunoassays such as the radio immunoassays (RIA) and the enzyme-linked immunosorbent assays (ELISA) are techniques for quantifying biochemical substances in specific media. The commonly used designs for these assays are those for classical calibration although this method has not been proven to be the best for the assays. One competing alternative calibration method that is rarely used in immunoassays is the parallel fitting method. In this paper we use the ELISA example to show that the classical calibration method is inaccurate and imprecise for predicting extreme unknown concentrations of biochemical substances whereas the parallel fitting method is reasonably accurate and precise for predicting both moderate and extreme unknown concentrations of the substances. Furthermore, we show how to extend optimal designs for classical calibration to optimal designs for parallel fitting. The conclusion of this paper is that the parallel fitting method should be preferred over the classical calibration method if in the area of application of the immunoassay both moderate and extreme unknown concentrations of the biochemical substance in the medium are of interest.