Method for the analytical determination of the concentration of a

Abstract

Method for the analytical determination of the concentration of a component of a medical sample, in which a reaction of the sample with reagents leads to a time-dependent change S(t) in a measured quantity S and the concentration C correlates according to an evaluation curve C(X) with an input variable X derived from S(t), in which the calibration curve is ambiguous for at least a portion of the possible X values. In order to assign an input variable X to one of the sub-sections and thereby to obtain an unambiguous correlation to a particular concentration C, a training run and an analysis run are performed. In the training run, a discrimination algorithm is performed at least once, in which a discriminator set is generated from measurements of S(t), a score is generated in each case from the latter with a multivariate statistical technique and it is checked whether the scores can be divided into separate subsets, in which the concentrations are correctly assigned to the sub-sections of the calibration curve. The discrimination algorithm is repeated with different discriminator sets generated according to different generation methods from the measurements of S(t), in order to establish an operative discrimination algorithm. In the analysis run, an analysis score is calculated from measurements of S(t) according to the operative discrimination algorithm, the analysis score is compared with the boundary score and the measurement result is assigned to one of the sub-curves of the calibration curve by comparison of the analysis score with the boundary score.