Valid: A generally applicable expert system-based program for the quantitative validation of the results of analytical methods

Abstract

Wolters, R., Van den Broek, A.C.M. and Kateman, G., 1990. VALID: A generally applicable expert system-based program for the quantitative validation of the results of analytical methods. Chemometrics and Intelligent Laboratory System, 9: 143–175. A system has been developed with which a quantitative validation of the results of an analysis can be achieved, for a specific application of an analytical method. The validation system, VALID, consist of an expert system-based program, in which knowledge about method validation is linked to the necessary (calculation) procedures and (mostly statistical) tests. The knowledge is represented in the system by means of a set of rules. The control of VALID takes place by a consultation of this rule set. Using this means of control renders the system suitable for a wide variety of analytical methods. There is no dependence on any particular analytical technique. During the course of a VALID run the reliability of the analysis system is estimated by means of the criterion of maximum total error, which incorporates both the random error and the systematic error of the results. Furthermore, a number of error sources are checked, such as the calibration procedure, interferences, matrix effects, and drift. The program provides the user with experimental designs and interprets the results of the experiments. Finally, the estimated reliability of the analytical results is compared to the demands specified by the user, which results in a judgement on the suitability of the technique. The performance of the validation system is illustrated by its application to the determination of cadmium in soil samples by electrothermal atomic absorption spectrometry.