Information theory and systematic toxicological analysis in “general unknown” poisoning cases

Abstract

 If qualitative analysis is aimed to detect large, imprecisely confined groups comprising hundreds or thousands of compounds – like during the analysis of questionable poisoning cases without direct hints, the so called “general unknowns”–, the evaluation of the identification certainty becomes equally important like the associated estimation of the necessary information capacity of the analytical procedures to be applied [1–4]. On this basis, the authors have outlined factors, which influence the usable information provided by analytical principles or procedures depending of the number and types (analytical categories, differences and similarities, detectability) of the analytes to be detected or included. A correlation between the usable information (excess or deficiency) and the probability of failing identifications is defined, and the information loss by interfering analytical signals is discussed. The concepts of discrimination power DP, identification power IP, and mean list length MLL [4–7] are associated with these considerations. Examples of analytical tasks in this “general unknown” environment of toxicological analysis are presented.