Fuzzy clustering analysis of the first 10 MEIC chemicals

In this paper, we discuss the classification results of the toxicological r esponses of 32 in vivo and in vitro test systems to the first 10 MEIC chemi cals. In this order we have used different fuzzy clustering algorithms, nam ely hierarchical fuzzy clustering, hierarchical and horizontal fuzzy charac teristics clustering and a new clustering technique, namely fuzzy hierarchi cal cross-classification. The characteristics clustering technique produces fuzzy partitions of the characteristics (chemicals) involved and thus it i s a useful tool for studying the (dis)similarities between different chemic als and for essential chemicals selection. The cross-classification algorit hm produces not only a fuzzy partition of the test systems analyzed, but al so a fuzzy partition of the considered 10 MEIC (multicentre evaluation of i n vitro cytotoxicity) chemicals. In this way it is possible to identify whi ch chemicals are responsible for the similarities or differences observed b etween different groups of test systems. Tn another way, there is a specifi c sensitivity of a chemical for one or more toxicological tests. (C) 1999 E lsevier Science Ltd. All rights reserved.