Calculating unknown concentrations from nonlinear standard curves

We describe an algorithm and program for calculating unknown concentration values from nonlinear standard data. The program was developed as an alternative to current data-handling techniques used for therapeutic drug enzyme immunoassay (EMIT) systems. The shape of the standard curve for a given analyte is required to be stable through the lifetime of a reagent lot. The relationship between the net change in milliabsorbance and the natural log of standard concentration is linearized once for each lot of reagents. For each subsequent time the assay is performed, one corrects the individual standard points by using the linearized relationship. Unknown concentrations are then calculated by using quadratic Lagrangian interpolation. The algorithm is faster and simpler than curve-fitting by nonlinear least squares, gives comparable results for unknowns, and rejects fewer assays for poor curve fit.