An automated titration system

Nevertheless there are two fundamentally different ways [2] of identifying sets of data. First, the cumulative collection of test results may be checked for plausibility. Mislabelled [3] results will be recognised if they differ greatly from those recorded earlier. The method, however, is not very effective as Sheiner et al [16] have shown. A Delta check method, [4] controlled by a computer, discovered only 50% of mis-[5] labelled specimens. On the other hand only 10% of the data [6] suspected of being mislabelled were in fact wrongly [7] identified. The second approach is based on the biochemical [8] individuality characterising a healthy, or a sick person unequivocally, if a sufficient number of tests is performed. [9] Young and his co-workers [17] have recently demonstrated [101 that the identity is reflected by multivariate data of biochemical test profiles. Profiles of individual samples were graphically displayed as computer-drawn faces and they [11] conclude from their experience the following: "Discriminant functions can be used to detect mislabelled specimens in [12] research studies and potentially in the clinical laboratory to detect misidentified samples from patients". Obviously this will not be a practicable method for sample identification in [13] our laboratories. But reflecting on the coherence of individuality and identification may be worth while.