[Methods of selection of cutoff points in the development and interpretation of diagnostic tests].

Abstract

Problems related to the optimization of diagnosis making are scrutinized. Applied immunological tests for diagnosis of ovarian cancers are used as an example of ROC curve calculation. Moreover, sensitivity and specificity grades are computed in order to obtain the optimum of diagnostical robustness. The ROC analysis is supplemented with application of Bayes diagnostical algorithm. The analysis is given also of other problems concerning with implementation of quantitative characteristics in the course of diagnostical decision making.